Case_Data_Master_Santander_v2.ipynb

{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.7.10","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"Este documento se encontra no repositório público \nhttps://github.com/obabilonia/case_santander_data_master\n\nIntrodução:\nEste trabalho foi definido na segunda fase do programa de certificação Data-Master do banco Santander Brasil. Nesta fase foi definido o desenvolvimento de um case pratico que consiste de 3 tarefas:\n<br>\n\n1. A primeira tarefa deste case é maximizar o lucro esperado por cliente considerando o conceito de identificação e retenção de clientes insatisfeitos.\n2. A segunda tarefa consiste em dar uma nota de 1 a 5 para cada cliente da base teste, respeitando a variável ‘TARGET’, isto é, o seu nível de satisfação, sendo 1 o mais insatisfeito e 5 o mais satisfeito. Ao dar essa nota deve-se ter em mente que somente os clientes com nota 1 serão alvos de uma ação de retenção e que o objetivo dessa ação é maximizar o lucro esperado por cliente (usando os mesmos valores da primeira questão).\n3. Todo conjunto de dados é passível de ser dividido em grupos coesos, conhecidos como agrupamentos naturais. A terceira tarefa é encontrar os três grupos naturais que possuem os maiores lucros esperados por cliente (usando os mesmos valores da primeira questão).\n\nCabe ressaltar que o case definido é um desafio Kaggle, e com uma exploração superficial no próprio desafio e em repositórios públicos é fácil achar diferentes soluções já propostas (com os respectivos códigos). Como é o caso do trabalho apresentado por Caio Martins (CaioMar) https://github.com/CaioMar/certificao_data_masters/blob/master/Certificacao%20Data%20Masters.ipynb\n<br>\nPor isso, neste trabalho o foco principal não será apresentar uma solução já encontrada, e sim a apresentação de uma abordagem diferente das já publicadas, tentando  apresentar novos insights utilizando os trabalhos prévios como base.\nTambém ressaltamos que na prática, não aplicariamos o passo a passo aqui empregado dado que entendemos que alguns dos métodos não são os mais eficiente, aqui o objetivo é mais didático ao invés de prático.  \n<br>\nResumo:\nAs abordagens apresentadas são definidas conforme segue:\n<br>\n#### Parte 1:\n1. Exploração e comentários sobre a variável resposta.\n2. Tratamento 1: Remoção de features com variância zero\n3. Tratamento 2: Remoção de features duplicadas\n4. Split do dataset em teste e treino\n5. Seleção do modelo RandomForest e Tuning de Hiperparametros\n6. Treinamento e teste do primeiro modelo\n7. Tratamento 3: Remoção de features com variância próxima a zero\n8. Tratamento 4: Remoção de features quase duplicadas (altamente correlacionadas)\n9. Retreinamento e teste do modelo\n\n#### Parte 2:\n1. Utilização do método predict_proba(), e apresentação da histograma de probabilidades\n2. Calculo e representação do ROC e AUC\n3. Definição de grupos de clientes segundo probabilidade de ser clientes insatisfeitos\n\n#### Parte 3:\n1. Reescalhamento das features\n2. Aplicação do K-means\n3. Representação gráfica do Elbow curve e Silhouttes\n4. Escolha do melhor K e segmentação dos clientes\n5. Avaliação de resultados\n\n#### Extra:\n1. Tratamento e teste de variáveis categóricas\n2. Aplicação e teste de métodos Undersampling e Oversampling\n3. Referencias\n","metadata":{}},{"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load\n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n\n# Input data files are available in the read-only \"../input/\" directory\n# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n\nimport os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))\n\n# You can write up to 20GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2021-09-03T11:34:16.403849Z","iopub.execute_input":"2021-09-03T11:34:16.404396Z","iopub.status.idle":"2021-09-03T11:34:16.419202Z","shell.execute_reply.started":"2021-09-03T11:34:16.404292Z","shell.execute_reply":"2021-09-03T11:34:16.418274Z"},"trusted":true},"execution_count":1,"outputs":[{"name":"stdout","text":"/kaggle/input/santander-customer-satisfaction/sample_submission.csv\n/kaggle/input/santander-customer-satisfaction/train.csv\n/kaggle/input/santander-customer-satisfaction/test.csv\n","output_type":"stream"}]},{"cell_type":"code","source":"#pd.read_csv('/kaggle/input/santander-customer-satisfaction/sample_submission.csv')","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:18.313575Z","iopub.execute_input":"2021-09-03T11:34:18.313984Z","iopub.status.idle":"2021-09-03T11:34:18.318205Z","shell.execute_reply.started":"2021-09-03T11:34:18.313952Z","shell.execute_reply":"2021-09-03T11:34:18.316909Z"},"trusted":true},"execution_count":2,"outputs":[]},{"cell_type":"code","source":"import numpy as np\nimport pandas as pd\nimport matplotlib\nimport matplotlib.pyplot as plt\nimport seaborn as sns\nsns.set_style('whitegrid')\nsns.set_context('notebook')\nplt.style.use('ggplot')\n\n%matplotlib inline\n\nfrom sklearn.metrics import pairwise_distances_argmin\nfrom sklearn.metrics import euclidean_distances\nfrom sklearn.tree import DecisionTreeClassifier\n\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.ensemble import GradientBoostingClassifier\nfrom sklearn.ensemble import AdaBoostClassifier\nfrom sklearn.neighbors import KNeighborsClassifier\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.naive_bayes import GaussianNB\nfrom sklearn.svm import SVC\n\n\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.model_selection import GridSearchCV\nfrom sklearn.model_selection import ShuffleSplit\nfrom sklearn.metrics import make_scorer, f1_score, fbeta_score, roc_auc_score, auc, confusion_matrix\nfrom sklearn.model_selection import StratifiedKFold\nfrom sklearn.feature_selection import VarianceThreshold\n\nfrom time import time\n\nimport warnings\nwarnings.filterwarnings(\"ignore\")\n\npd.set_option('display.max_columns', None)\npd.set_option('display.max_rows', None)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:18.738162Z","iopub.execute_input":"2021-09-03T11:34:18.738732Z","iopub.status.idle":"2021-09-03T11:34:20.026064Z","shell.execute_reply.started":"2021-09-03T11:34:18.738667Z","shell.execute_reply":"2021-09-03T11:34:20.025357Z"},"trusted":true},"execution_count":3,"outputs":[]},{"cell_type":"code","source":"def see_categories(data, target):\n    count = pd.DataFrame(data[target].value_counts())\n    percent = pd.DataFrame(data[target].value_counts(normalize=True)*100)\n    table = pd.concat([count, percent], axis=1)\n    table.columns = ['qtd', 'perc']\n    return table.sort_values('perc', ascending=False)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:20.842569Z","iopub.execute_input":"2021-09-03T11:34:20.843070Z","iopub.status.idle":"2021-09-03T11:34:20.848738Z","shell.execute_reply.started":"2021-09-03T11:34:20.843038Z","shell.execute_reply":"2021-09-03T11:34:20.847852Z"},"trusted":true},"execution_count":4,"outputs":[]},{"cell_type":"code","source":"def corr_heatmap (data, figsize_1= 10, figsize_2 = 10, see_numbers= False):\n    fig, plot = plt.subplots(figsize=(figsize_1, figsize_2)) \n    plot = sns.heatmap(pd.DataFrame.corr(data), linewidths=0.3, annot = see_numbers)\n    return plot","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:21.591246Z","iopub.execute_input":"2021-09-03T11:34:21.591808Z","iopub.status.idle":"2021-09-03T11:34:21.611344Z","shell.execute_reply.started":"2021-09-03T11:34:21.591757Z","shell.execute_reply":"2021-09-03T11:34:21.610439Z"},"trusted":true},"execution_count":5,"outputs":[]},{"cell_type":"code","source":"data = pd.read_csv('/kaggle/input/santander-customer-satisfaction/train.csv')\ndf_test = pd.read_csv('/kaggle/input/santander-customer-satisfaction/test.csv')\n\nprint(data.shape, df_test.shape) ","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:23.932042Z","iopub.execute_input":"2021-09-03T11:34:23.934131Z","iopub.status.idle":"2021-09-03T11:34:29.275081Z","shell.execute_reply.started":"2021-09-03T11:34:23.934094Z","shell.execute_reply":"2021-09-03T11:34:29.274079Z"},"trusted":true},"execution_count":6,"outputs":[{"name":"stdout","text":"(76020, 371) (75818, 370)\n","output_type":"stream"}]},{"cell_type":"code","source":"data.head()","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:29.277549Z","iopub.execute_input":"2021-09-03T11:34:29.278156Z","iopub.status.idle":"2021-09-03T11:34:29.543233Z","shell.execute_reply.started":"2021-09-03T11:34:29.278107Z","shell.execute_reply":"2021-09-03T11:34:29.542094Z"},"trusted":true},"execution_count":7,"outputs":[{"execution_count":7,"output_type":"execute_result","data":{"text/plain":"   ID  var3  var15  imp_ent_var16_ult1  imp_op_var39_comer_ult1  \\\n0   1     2     23                 0.0                      0.0   \n1   3     2     34                 0.0                      0.0   \n2   4     2     23                 0.0                      0.0   \n3   8     2     37                 0.0                    195.0   \n4  10     2     39                 0.0                      0.0   \n\n   imp_op_var39_comer_ult3  imp_op_var40_comer_ult1  imp_op_var40_comer_ult3  \\\n0                      0.0                      0.0                      0.0   \n1                      0.0                      0.0                      0.0   \n2                      0.0                      0.0                      0.0   \n3                    195.0                      0.0                      0.0   \n4                      0.0                      0.0                      0.0   \n\n   imp_op_var40_efect_ult1  imp_op_var40_efect_ult3  imp_op_var40_ult1  \\\n0                      0.0                      0.0                0.0   \n1                      0.0                      0.0                0.0   \n2                      0.0                      0.0                0.0   \n3                      0.0                      0.0                0.0   \n4                      0.0                      0.0                0.0   \n\n   imp_op_var41_comer_ult1  imp_op_var41_comer_ult3  imp_op_var41_efect_ult1  \\\n0                      0.0                      0.0                      0.0   \n1                      0.0                      0.0                      0.0   \n2                      0.0                      0.0                      0.0   \n3                    195.0                    195.0                      0.0   \n4                      0.0                      0.0                      0.0   \n\n   imp_op_var41_efect_ult3  imp_op_var41_ult1  imp_op_var39_efect_ult1  \\\n0                      0.0                0.0                      0.0   \n1                      0.0                0.0                      0.0   \n2                      0.0                0.0                      0.0   \n3                      0.0              195.0                      0.0   \n4                      0.0                0.0                      0.0   \n\n   imp_op_var39_efect_ult3  imp_op_var39_ult1  imp_sal_var16_ult1  ind_var1_0  \\\n0                      0.0                0.0                 0.0           0   \n1                      0.0                0.0                 0.0           0   \n2                      0.0                0.0                 0.0           0   \n3                      0.0              195.0                 0.0           0   \n4                      0.0                0.0                 0.0           0   \n\n   ind_var1  ind_var2_0  ind_var2  ind_var5_0  ind_var5  ind_var6_0  ind_var6  \\\n0         0           0         0           1         0           0         0   \n1         0           0         0           1         0           0         0   \n2         0           0         0           1         1           0         0   \n3         0           0         0           1         1           0         0   \n4         0           0         0           1         0           0         0   \n\n   ind_var8_0  ind_var8  ind_var12_0  ind_var12  ind_var13_0  \\\n0           0         0            0          0            0   \n1           0         0            0          0            1   \n2           0         0            0          0            0   \n3           0         0            0          0            0   \n4           0         0            1          1            0   \n\n   ind_var13_corto_0  ind_var13_corto  ind_var13_largo_0  ind_var13_largo  \\\n0                  0                0                  0                0   \n1                  1                1                  0                0   \n2                  0                0                  0                0   \n3                  0                0                  0                0   \n4                  0                0                  0                0   \n\n   ind_var13_medio_0  ind_var13_medio  ind_var13  ind_var14_0  ind_var14  \\\n0                  0                0          0            0          0   \n1                  0                0          1            0          0   \n2                  0                0          0            0          0   \n3                  0                0          0            0          0   \n4                  0                0          0            0          0   \n\n   ind_var17_0  ind_var17  ind_var18_0  ind_var18  ind_var19  ind_var20_0  \\\n0            0          0            0          0          0            0   \n1            0          0            0          0          0            0   \n2            0          0            0          0          0            0   \n3            0          0            0          0          0            0   \n4            0          0            0          0          0            0   \n\n   ind_var20  ind_var24_0  ind_var24  ind_var25_cte  ind_var26_0  \\\n0          0            0          0              0            0   \n1          0            0          0              0            0   \n2          0            0          0              0            0   \n3          0            0          0              0            0   \n4          0            1          1              0            0   \n\n   ind_var26_cte  ind_var26  ind_var25_0  ind_var25  ind_var27_0  ind_var28_0  \\\n0              0          0            0          0            0            0   \n1              0          0            0          0            0            0   \n2              0          0            0          0            0            0   \n3              0          0            0          0            0            0   \n4              0          0            0          0            0            0   \n\n   ind_var28  ind_var27  ind_var29_0  ind_var29  ind_var30_0  ind_var30  \\\n0          0          0            0          0            1          0   \n1          0          0            0          0            1          1   \n2          0          0            0          0            1          1   \n3          0          0            0          0            1          1   \n4          0          0            0          0            1          1   \n\n   ind_var31_0  ind_var31  ind_var32_cte  ind_var32_0  ind_var32  ind_var33_0  \\\n0            0          0              0            0          0            0   \n1            0          0              0            0          0            0   \n2            0          0              0            0          0            0   \n3            0          0              0            0          0            0   \n4            0          0              0            0          0            0   \n\n   ind_var33  ind_var34_0  ind_var34  ind_var37_cte  ind_var37_0  ind_var37  \\\n0          0            0          0              0            0          0   \n1          0            0          0              0            0          0   \n2          0            0          0              0            0          0   \n3          0            0          0              1            1          1   \n4          0            0          0              0            0          0   \n\n   ind_var39_0  ind_var40_0  ind_var40  ind_var41_0  ind_var41  ind_var39  \\\n0            1            0          0            1          0          0   \n1            1            0          0            1          0          0   \n2            1            0          0            1          0          0   \n3            1            0          0            1          0          0   \n4            1            0          0            1          0          0   \n\n   ind_var44_0  ind_var44  ind_var46_0  ind_var46  num_var1_0  num_var1  \\\n0            0          0            0          0           0         0   \n1            0          0            0          0           0         0   \n2            0          0            0          0           0         0   \n3            0          0            0          0           0         0   \n4            0          0            0          0           0         0   \n\n   num_var4  num_var5_0  num_var5  num_var6_0  num_var6  num_var8_0  num_var8  \\\n0         0           3         0           0         0           0         0   \n1         1           3         0           0         0           0         0   \n2         1           3         3           0         0           0         0   \n3         3           3         3           0         0           0         0   \n4         1           3         0           0         0           0         0   \n\n   num_var12_0  num_var12  num_var13_0  num_var13_corto_0  num_var13_corto  \\\n0            0          0            0                  0                0   \n1            0          0            3                  3                3   \n2            0          0            0                  0                0   \n3            0          0            0                  0                0   \n4            3          3            0                  0                0   \n\n   num_var13_largo_0  num_var13_largo  num_var13_medio_0  num_var13_medio  \\\n0                  0                0                  0                0   \n1                  0                0                  0                0   \n2                  0                0                  0                0   \n3                  0                0                  0                0   \n4                  0                0                  0                0   \n\n   num_var13  num_var14_0  num_var14  num_var17_0  num_var17  num_var18_0  \\\n0          0            0          0            0          0            0   \n1          3            0          0            0          0            0   \n2          0            0          0            0          0            0   \n3          0            0          0            0          0            0   \n4          0            0          0            0          0            0   \n\n   num_var18  num_var20_0  num_var20  num_var24_0  num_var24  num_var26_0  \\\n0          0            0          0            0          0            0   \n1          0            0          0            0          0            0   \n2          0            0          0            0          0            0   \n3          0            0          0            0          0            0   \n4          0            0          0            3          3            0   \n\n   num_var26  num_var25_0  num_var25  num_op_var40_hace2  num_op_var40_hace3  \\\n0          0            0          0                   0                   0   \n1          0            0          0                   0                   0   \n2          0            0          0                   0                   0   \n3          0            0          0                   0                   0   \n4          0            0          0                   0                   0   \n\n   num_op_var40_ult1  num_op_var40_ult3  num_op_var41_hace2  \\\n0                  0                  0                   0   \n1                  0                  0                   0   \n2                  0                  0                   0   \n3                  0                  0                   0   \n4                  0                  0                   0   \n\n   num_op_var41_hace3  num_op_var41_ult1  num_op_var41_ult3  \\\n0                   0                  0                  0   \n1                   0                  0                  0   \n2                   0                  0                  0   \n3                   0                  9                  9   \n4                   0                  0                  0   \n\n   num_op_var39_hace2  num_op_var39_hace3  num_op_var39_ult1  \\\n0                   0                   0                  0   \n1                   0                   0                  0   \n2                   0                   0                  0   \n3                   0                   0                  9   \n4                   0                   0                  0   \n\n   num_op_var39_ult3  num_var27_0  num_var28_0  num_var28  num_var27  \\\n0                  0            0            0          0          0   \n1                  0            0            0          0          0   \n2                  0            0            0          0          0   \n3                  9            0            0          0          0   \n4                  0            0            0          0          0   \n\n   num_var29_0  num_var29  num_var30_0  num_var30  num_var31_0  num_var31  \\\n0            0          0            3          0            0          0   \n1            0          0            6          3            0          0   \n2            0          0            3          3            0          0   \n3            0          0            3          3            0          0   \n4            0          0            6          3            0          0   \n\n   num_var32_0  num_var32  num_var33_0  num_var33  num_var34_0  num_var34  \\\n0            0          0            0          0            0          0   \n1            0          0            0          0            0          0   \n2            0          0            0          0            0          0   \n3            0          0            0          0            0          0   \n4            0          0            0          0            0          0   \n\n   num_var35  num_var37_med_ult2  num_var37_0  num_var37  num_var39_0  \\\n0          0                   0            0          0            3   \n1          3                   0            0          0            3   \n2          3                   0            0          0            3   \n3          9                   6            6          6            3   \n4          3                   0            0          0            3   \n\n   num_var40_0  num_var40  num_var41_0  num_var41  num_var39  num_var42_0  \\\n0            0          0            3          0          0            3   \n1            0          0            3          0          0            3   \n2            0          0            3          0          0            3   \n3            0          0            3          0          0            3   \n4            0          0            3          0          0            6   \n\n   num_var42  num_var44_0  num_var44  num_var46_0  num_var46  saldo_var1  \\\n0          0            0          0            0          0         0.0   \n1          0            0          0            0          0         0.0   \n2          3            0          0            0          0         0.0   \n3          3            0          0            0          0         0.0   \n4          3            0          0            0          0         0.0   \n\n   saldo_var5  saldo_var6  saldo_var8  saldo_var12  saldo_var13_corto  \\\n0        0.00         0.0         0.0          0.0                0.0   \n1        0.00         0.0         0.0          0.0              300.0   \n2        3.00         0.0         0.0          0.0                0.0   \n3       70.62         0.0         0.0          0.0                0.0   \n4        0.00         0.0         0.0     135003.0                0.0   \n\n   saldo_var13_largo  saldo_var13_medio  saldo_var13  saldo_var14  \\\n0                0.0                  0          0.0          0.0   \n1                0.0                  0        300.0          0.0   \n2                0.0                  0          0.0          0.0   \n3                0.0                  0          0.0          0.0   \n4                0.0                  0          0.0          0.0   \n\n   saldo_var17  saldo_var18  saldo_var20  saldo_var24  saldo_var26  \\\n0          0.0            0          0.0          0.0          0.0   \n1          0.0            0          0.0          0.0          0.0   \n2          0.0            0          0.0          0.0          0.0   \n3          0.0            0          0.0          0.0          0.0   \n4          0.0            0          0.0     135003.0          0.0   \n\n   saldo_var25  saldo_var28  saldo_var27  saldo_var29  saldo_var30  \\\n0          0.0            0            0          0.0         0.00   \n1          0.0            0            0          0.0       300.00   \n2          0.0            0            0          0.0         3.00   \n3          0.0            0            0          0.0        70.62   \n4          0.0            0            0          0.0    135003.00   \n\n   saldo_var31  saldo_var32  saldo_var33  saldo_var34  saldo_var37  \\\n0          0.0          0.0          0.0            0         0.00   \n1          0.0          0.0          0.0            0         0.00   \n2          0.0          0.0          0.0            0         0.00   \n3          0.0          0.0          0.0            0        34.95   \n4          0.0          0.0          0.0            0         0.00   \n\n   saldo_var40  saldo_var41  saldo_var42  saldo_var44  saldo_var46  var36  \\\n0          0.0            0         0.00          0.0            0     99   \n1          0.0            0         0.00          0.0            0      3   \n2          0.0            0         3.00          0.0            0     99   \n3          0.0            0        70.62          0.0            0      2   \n4          0.0            0    135003.00          0.0            0      1   \n\n   delta_imp_amort_var18_1y3  delta_imp_amort_var34_1y3  \\\n0                          0                          0   \n1                          0                          0   \n2                          0                          0   \n3                          0                          0   \n4                          0                          0   \n\n   delta_imp_aport_var13_1y3  delta_imp_aport_var17_1y3  \\\n0                        0.0                        0.0   \n1                       -1.0                        0.0   \n2                        0.0                        0.0   \n3                        0.0                        0.0   \n4                        0.0                        0.0   \n\n   delta_imp_aport_var33_1y3  delta_imp_compra_var44_1y3  \\\n0                        0.0                         0.0   \n1                        0.0                         0.0   \n2                        0.0                         0.0   \n3                        0.0                         0.0   \n4                        0.0                         0.0   \n\n   delta_imp_reemb_var13_1y3  delta_imp_reemb_var17_1y3  \\\n0                          0                          0   \n1                          0                          0   \n2                          0                          0   \n3                          0                          0   \n4                          0                          0   \n\n   delta_imp_reemb_var33_1y3  delta_imp_trasp_var17_in_1y3  \\\n0                          0                             0   \n1                          0                             0   \n2                          0                             0   \n3                          0                             0   \n4                          0                             0   \n\n   delta_imp_trasp_var17_out_1y3  delta_imp_trasp_var33_in_1y3  \\\n0                              0                             0   \n1                              0                             0   \n2                              0                             0   \n3                              0                             0   \n4                              0                             0   \n\n   delta_imp_trasp_var33_out_1y3  delta_imp_venta_var44_1y3  \\\n0                              0                        0.0   \n1                              0                        0.0   \n2                              0                        0.0   \n3                              0                        0.0   \n4                              0                        0.0   \n\n   delta_num_aport_var13_1y3  delta_num_aport_var17_1y3  \\\n0                        0.0                        0.0   \n1                       -1.0                        0.0   \n2                        0.0                        0.0   \n3                        0.0                        0.0   \n4                        0.0                        0.0   \n\n   delta_num_aport_var33_1y3  delta_num_compra_var44_1y3  \\\n0                        0.0                         0.0   \n1                        0.0                         0.0   \n2                        0.0                         0.0   \n3                        0.0                         0.0   \n4                        0.0                         0.0   \n\n   delta_num_reemb_var13_1y3  delta_num_reemb_var17_1y3  \\\n0                          0                          0   \n1                          0                          0   \n2                          0                          0   \n3                          0                          0   \n4                          0                          0   \n\n   delta_num_reemb_var33_1y3  delta_num_trasp_var17_in_1y3  \\\n0                          0                             0   \n1                          0                             0   \n2                          0                             0   \n3                          0                             0   \n4                          0                             0   \n\n   delta_num_trasp_var17_out_1y3  delta_num_trasp_var33_in_1y3  \\\n0                              0                             0   \n1                              0                             0   \n2                              0                             0   \n3                              0                             0   \n4                              0                             0   \n\n   delta_num_trasp_var33_out_1y3  delta_num_venta_var44_1y3  \\\n0                              0                        0.0   \n1                              0                        0.0   \n2                              0                        0.0   \n3                              0                        0.0   \n4                              0                        0.0   \n\n   imp_amort_var18_hace3  imp_amort_var18_ult1  imp_amort_var34_hace3  \\\n0                      0                   0.0                      0   \n1                      0                   0.0                      0   \n2                      0                   0.0                      0   \n3                      0                   0.0                      0   \n4                      0                   0.0                      0   \n\n   imp_amort_var34_ult1  imp_aport_var13_hace3  imp_aport_var13_ult1  \\\n0                   0.0                    0.0                   0.0   \n1                   0.0                  300.0                   0.0   \n2                   0.0                    0.0                   0.0   \n3                   0.0                    0.0                   0.0   \n4                   0.0                    0.0                   0.0   \n\n   imp_aport_var17_hace3  imp_aport_var17_ult1  imp_aport_var33_hace3  \\\n0                    0.0                   0.0                      0   \n1                    0.0                   0.0                      0   \n2                    0.0                   0.0                      0   \n3                    0.0                   0.0                      0   \n4                    0.0                   0.0                      0   \n\n   imp_aport_var33_ult1  imp_var7_emit_ult1  imp_var7_recib_ult1  \\\n0                     0                 0.0                  0.0   \n1                     0                 0.0                  0.0   \n2                     0                 0.0                  0.0   \n3                     0                 0.0                  0.0   \n4                     0                 0.0                  0.0   \n\n   imp_compra_var44_hace3  imp_compra_var44_ult1  imp_reemb_var13_hace3  \\\n0                     0.0                    0.0                      0   \n1                     0.0                    0.0                      0   \n2                     0.0                    0.0                      0   \n3                     0.0                    0.0                      0   \n4                     0.0                    0.0                      0   \n\n   imp_reemb_var13_ult1  imp_reemb_var17_hace3  imp_reemb_var17_ult1  \\\n0                   0.0                    0.0                   0.0   \n1                   0.0                    0.0                   0.0   \n2                   0.0                    0.0                   0.0   \n3                   0.0                    0.0                   0.0   \n4                   0.0                    0.0                   0.0   \n\n   imp_reemb_var33_hace3  imp_reemb_var33_ult1  imp_var43_emit_ult1  \\\n0                      0                     0                  0.0   \n1                      0                     0                  0.0   \n2                      0                     0                  0.0   \n3                      0                     0                  0.0   \n4                      0                     0             135003.0   \n\n   imp_trans_var37_ult1  imp_trasp_var17_in_hace3  imp_trasp_var17_in_ult1  \\\n0                   0.0                       0.0                      0.0   \n1                   0.0                       0.0                      0.0   \n2                   0.0                       0.0                      0.0   \n3                   0.0                       0.0                      0.0   \n4              270003.0                       0.0                      0.0   \n\n   imp_trasp_var17_out_hace3  imp_trasp_var17_out_ult1  \\\n0                          0                       0.0   \n1                          0                       0.0   \n2                          0                       0.0   \n3                          0                       0.0   \n4                          0                       0.0   \n\n   imp_trasp_var33_in_hace3  imp_trasp_var33_in_ult1  \\\n0                       0.0                      0.0   \n1                       0.0                      0.0   \n2                       0.0                      0.0   \n3                       0.0                      0.0   \n4                       0.0                      0.0   \n\n   imp_trasp_var33_out_hace3  imp_trasp_var33_out_ult1  imp_venta_var44_hace3  \\\n0                          0                         0                    0.0   \n1                          0                         0                    0.0   \n2                          0                         0                    0.0   \n3                          0                         0                    0.0   \n4                          0                         0                    0.0   \n\n   imp_venta_var44_ult1  ind_var7_emit_ult1  ind_var7_recib_ult1  \\\n0                   0.0                   0                    0   \n1                   0.0                   0                    0   \n2                   0.0                   0                    0   \n3                   0.0                   0                    0   \n4                   0.0                   0                    0   \n\n   ind_var10_ult1  ind_var10cte_ult1  ind_var9_cte_ult1  ind_var9_ult1  \\\n0               0                  0                  0              0   \n1               0                  0                  0              0   \n2               0                  0                  0              0   \n3               1                  1                  1              1   \n4               0                  0                  0              0   \n\n   ind_var43_emit_ult1  ind_var43_recib_ult1  var21  num_var2_0_ult1  \\\n0                    0                     0      0                0   \n1                    0                     0      0                0   \n2                    0                     0      0                0   \n3                    0                     0      0                0   \n4                    1                     1      0                0   \n\n   num_var2_ult1  num_aport_var13_hace3  num_aport_var13_ult1  \\\n0              0                      0                     0   \n1              0                      3                     0   \n2              0                      0                     0   \n3              0                      0                     0   \n4              0                      0                     0   \n\n   num_aport_var17_hace3  num_aport_var17_ult1  num_aport_var33_hace3  \\\n0                      0                     0                      0   \n1                      0                     0                      0   \n2                      0                     0                      0   \n3                      0                     0                      0   \n4                      0                     0                      0   \n\n   num_aport_var33_ult1  num_var7_emit_ult1  num_var7_recib_ult1  \\\n0                     0                   0                    0   \n1                     0                   0                    0   \n2                     0                   0                    0   \n3                     0                   0                    0   \n4                     0                   0                    0   \n\n   num_compra_var44_hace3  num_compra_var44_ult1  num_ent_var16_ult1  \\\n0                       0                      0                   0   \n1                       0                      0                   0   \n2                       0                      0                   0   \n3                       0                      0                   0   \n4                       0                      0                   0   \n\n   num_var22_hace2  num_var22_hace3  num_var22_ult1  num_var22_ult3  \\\n0                0                0               0               0   \n1                0                0               0               0   \n2                0                0               0               0   \n3                3                0               0               3   \n4                3                0               6               9   \n\n   num_med_var22_ult3  num_med_var45_ult3  num_meses_var5_ult3  \\\n0                   0                   0                    0   \n1                   0                   0                    1   \n2                   0                   0                    3   \n3                   0                  15                    2   \n4                   3                   0                    3   \n\n   num_meses_var8_ult3  num_meses_var12_ult3  num_meses_var13_corto_ult3  \\\n0                    0                     0                           0   \n1                    0                     0                           3   \n2                    0                     0                           0   \n3                    0                     0                           0   \n4                    0                     1                           0   \n\n   num_meses_var13_largo_ult3  num_meses_var13_medio_ult3  \\\n0                           0                           0   \n1                           0                           0   \n2                           0                           0   \n3                           0                           0   \n4                           0                           0   \n\n   num_meses_var17_ult3  num_meses_var29_ult3  num_meses_var33_ult3  \\\n0                     0                     0                     0   \n1                     0                     0                     0   \n2                     0                     0                     0   \n3                     0                     0                     0   \n4                     0                     0                     0   \n\n   num_meses_var39_vig_ult3  num_meses_var44_ult3  num_op_var39_comer_ult1  \\\n0                         2                     0                        0   \n1                         2                     0                        0   \n2                         1                     0                        0   \n3                         1                     0                        9   \n4                         2                     0                        0   \n\n   num_op_var39_comer_ult3  num_op_var40_comer_ult1  num_op_var40_comer_ult3  \\\n0                        0                        0                        0   \n1                        0                        0                        0   \n2                        0                        0                        0   \n3                        9                        0                        0   \n4                        0                        0                        0   \n\n   num_op_var40_efect_ult1  num_op_var40_efect_ult3  num_op_var41_comer_ult1  \\\n0                        0                        0                        0   \n1                        0                        0                        0   \n2                        0                        0                        0   \n3                        0                        0                        9   \n4                        0                        0                        0   \n\n   num_op_var41_comer_ult3  num_op_var41_efect_ult1  num_op_var41_efect_ult3  \\\n0                        0                        0                        0   \n1                        0                        0                        0   \n2                        0                        0                        0   \n3                        9                        0                        0   \n4                        0                        0                        0   \n\n   num_op_var39_efect_ult1  num_op_var39_efect_ult3  num_reemb_var13_hace3  \\\n0                        0                        0                      0   \n1                        0                        0                      0   \n2                        0                        0                      0   \n3                        0                        0                      0   \n4                        0                        0                      0   \n\n   num_reemb_var13_ult1  num_reemb_var17_hace3  num_reemb_var17_ult1  \\\n0                     0                      0                     0   \n1                     0                      0                     0   \n2                     0                      0                     0   \n3                     0                      0                     0   \n4                     0                      0                     0   \n\n   num_reemb_var33_hace3  num_reemb_var33_ult1  num_sal_var16_ult1  \\\n0                      0                     0                   0   \n1                      0                     0                   0   \n2                      0                     0                   0   \n3                      0                     0                   0   \n4                      0                     0                   0   \n\n   num_var43_emit_ult1  num_var43_recib_ult1  num_trasp_var11_ult1  \\\n0                    0                     0                     0   \n1                    0                     0                     0   \n2                    0                     0                     0   \n3                    0                     0                     0   \n4                    3                     6                     3   \n\n   num_trasp_var17_in_hace3  num_trasp_var17_in_ult1  \\\n0                         0                        0   \n1                         0                        0   \n2                         0                        0   \n3                         0                        0   \n4                         0                        0   \n\n   num_trasp_var17_out_hace3  num_trasp_var17_out_ult1  \\\n0                          0                         0   \n1                          0                         0   \n2                          0                         0   \n3                          0                         0   \n4                          0                         0   \n\n   num_trasp_var33_in_hace3  num_trasp_var33_in_ult1  \\\n0                         0                        0   \n1                         0                        0   \n2                         0                        0   \n3                         0                        0   \n4                         0                        0   \n\n   num_trasp_var33_out_hace3  num_trasp_var33_out_ult1  num_venta_var44_hace3  \\\n0                          0                         0                      0   \n1                          0                         0                      0   \n2                          0                         0                      0   \n3                          0                         0                      0   \n4                          0                         0                      0   \n\n   num_venta_var44_ult1  num_var45_hace2  num_var45_hace3  num_var45_ult1  \\\n0                     0                0                0               0   \n1                     0                0                0               0   \n2                     0                0                0               0   \n3                     0               27                3              18   \n4                     0                0                0               0   \n\n   num_var45_ult3  saldo_var2_ult1  saldo_medio_var5_hace2  \\\n0               0                0                    0.00   \n1               0                0                    0.00   \n2               0                0                    3.00   \n3              48                0                  186.09   \n4               0                0                    3.00   \n\n   saldo_medio_var5_hace3  saldo_medio_var5_ult1  saldo_medio_var5_ult3  \\\n0                    0.00                   0.00                   0.00   \n1                   88.89                   0.00                   0.00   \n2                    0.18                   3.00                   2.07   \n3                    0.00                  91.56                 138.84   \n4                    0.30               40501.08               13501.47   \n\n   saldo_medio_var8_hace2  saldo_medio_var8_hace3  saldo_medio_var8_ult1  \\\n0                     0.0                     0.0                    0.0   \n1                     0.0                     0.0                    0.0   \n2                     0.0                     0.0                    0.0   \n3                     0.0                     0.0                    0.0   \n4                     0.0                     0.0                    0.0   \n\n   saldo_medio_var8_ult3  saldo_medio_var12_hace2  saldo_medio_var12_hace3  \\\n0                    0.0                      0.0                      0.0   \n1                    0.0                      0.0                      0.0   \n2                    0.0                      0.0                      0.0   \n3                    0.0                      0.0                      0.0   \n4                    0.0                      0.0                      0.0   \n\n   saldo_medio_var12_ult1  saldo_medio_var12_ult3  \\\n0                    0.00                    0.00   \n1                    0.00                    0.00   \n2                    0.00                    0.00   \n3                    0.00                    0.00   \n4                85501.89                85501.89   \n\n   saldo_medio_var13_corto_hace2  saldo_medio_var13_corto_hace3  \\\n0                            0.0                           0.00   \n1                          300.0                         122.22   \n2                            0.0                           0.00   \n3                            0.0                           0.00   \n4                            0.0                           0.00   \n\n   saldo_medio_var13_corto_ult1  saldo_medio_var13_corto_ult3  \\\n0                           0.0                          0.00   \n1                         300.0                        240.75   \n2                           0.0                          0.00   \n3                           0.0                          0.00   \n4                           0.0                          0.00   \n\n   saldo_medio_var13_largo_hace2  saldo_medio_var13_largo_hace3  \\\n0                            0.0                            0.0   \n1                            0.0                            0.0   \n2                            0.0                            0.0   \n3                            0.0                            0.0   \n4                            0.0                            0.0   \n\n   saldo_medio_var13_largo_ult1  saldo_medio_var13_largo_ult3  \\\n0                           0.0                           0.0   \n1                           0.0                           0.0   \n2                           0.0                           0.0   \n3                           0.0                           0.0   \n4                           0.0                           0.0   \n\n   saldo_medio_var13_medio_hace2  saldo_medio_var13_medio_hace3  \\\n0                            0.0                              0   \n1                            0.0                              0   \n2                            0.0                              0   \n3                            0.0                              0   \n4                            0.0                              0   \n\n   saldo_medio_var13_medio_ult1  saldo_medio_var13_medio_ult3  \\\n0                             0                           0.0   \n1                             0                           0.0   \n2                             0                           0.0   \n3                             0                           0.0   \n4                             0                           0.0   \n\n   saldo_medio_var17_hace2  saldo_medio_var17_hace3  saldo_medio_var17_ult1  \\\n0                      0.0                      0.0                     0.0   \n1                      0.0                      0.0                     0.0   \n2                      0.0                      0.0                     0.0   \n3                      0.0                      0.0                     0.0   \n4                      0.0                      0.0                     0.0   \n\n   saldo_medio_var17_ult3  saldo_medio_var29_hace2  saldo_medio_var29_hace3  \\\n0                     0.0                      0.0                      0.0   \n1                     0.0                      0.0                      0.0   \n2                     0.0                      0.0                      0.0   \n3                     0.0                      0.0                      0.0   \n4                     0.0                      0.0                      0.0   \n\n   saldo_medio_var29_ult1  saldo_medio_var29_ult3  saldo_medio_var33_hace2  \\\n0                     0.0                     0.0                      0.0   \n1                     0.0                     0.0                      0.0   \n2                     0.0                     0.0                      0.0   \n3                     0.0                     0.0                      0.0   \n4                     0.0                     0.0                      0.0   \n\n   saldo_medio_var33_hace3  saldo_medio_var33_ult1  saldo_medio_var33_ult3  \\\n0                      0.0                     0.0                     0.0   \n1                      0.0                     0.0                     0.0   \n2                      0.0                     0.0                     0.0   \n3                      0.0                     0.0                     0.0   \n4                      0.0                     0.0                     0.0   \n\n   saldo_medio_var44_hace2  saldo_medio_var44_hace3  saldo_medio_var44_ult1  \\\n0                      0.0                      0.0                     0.0   \n1                      0.0                      0.0                     0.0   \n2                      0.0                      0.0                     0.0   \n3                      0.0                      0.0                     0.0   \n4                      0.0                      0.0                     0.0   \n\n   saldo_medio_var44_ult3          var38  TARGET  \n0                     0.0   39205.170000       0  \n1                     0.0   49278.030000       0  \n2                     0.0   67333.770000       0  \n3                     0.0   64007.970000       0  \n4                     0.0  117310.979016       0  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>ID</th>\n      <th>var3</th>\n      <th>var15</th>\n      <th>imp_ent_var16_ult1</th>\n      <th>imp_op_var39_comer_ult1</th>\n      <th>imp_op_var39_comer_ult3</th>\n      <th>imp_op_var40_comer_ult1</th>\n      <th>imp_op_var40_comer_ult3</th>\n      <th>imp_op_var40_efect_ult1</th>\n      <th>imp_op_var40_efect_ult3</th>\n      <th>imp_op_var40_ult1</th>\n      <th>imp_op_var41_comer_ult1</th>\n      <th>imp_op_var41_comer_ult3</th>\n      <th>imp_op_var41_efect_ult1</th>\n      <th>imp_op_var41_efect_ult3</th>\n      <th>imp_op_var41_ult1</th>\n      <th>imp_op_var39_efect_ult1</th>\n      <th>imp_op_var39_efect_ult3</th>\n      <th>imp_op_var39_ult1</th>\n      <th>imp_sal_var16_ult1</th>\n      <th>ind_var1_0</th>\n      <th>ind_var1</th>\n      <th>ind_var2_0</th>\n      <th>ind_var2</th>\n      <th>ind_var5_0</th>\n      <th>ind_var5</th>\n      <th>ind_var6_0</th>\n      <th>ind_var6</th>\n      <th>ind_var8_0</th>\n      <th>ind_var8</th>\n      <th>ind_var12_0</th>\n      <th>ind_var12</th>\n      <th>ind_var13_0</th>\n      <th>ind_var13_corto_0</th>\n      <th>ind_var13_corto</th>\n      <th>ind_var13_largo_0</th>\n      <th>ind_var13_largo</th>\n      <th>ind_var13_medio_0</th>\n      <th>ind_var13_medio</th>\n      <th>ind_var13</th>\n      <th>ind_var14_0</th>\n      <th>ind_var14</th>\n      <th>ind_var17_0</th>\n      <th>ind_var17</th>\n      <th>ind_var18_0</th>\n      <th>ind_var18</th>\n      <th>ind_var19</th>\n      <th>ind_var20_0</th>\n      <th>ind_var20</th>\n      <th>ind_var24_0</th>\n      <th>ind_var24</th>\n      <th>ind_var25_cte</th>\n      <th>ind_var26_0</th>\n      <th>ind_var26_cte</th>\n      <th>ind_var26</th>\n      <th>ind_var25_0</th>\n      <th>ind_var25</th>\n      <th>ind_var27_0</th>\n      <th>ind_var28_0</th>\n      <th>ind_var28</th>\n      <th>ind_var27</th>\n      <th>ind_var29_0</th>\n      <th>ind_var29</th>\n      <th>ind_var30_0</th>\n      <th>ind_var30</th>\n      <th>ind_var31_0</th>\n      <th>ind_var31</th>\n      <th>ind_var32_cte</th>\n      <th>ind_var32_0</th>\n      <th>ind_var32</th>\n      <th>ind_var33_0</th>\n      <th>ind_var33</th>\n      <th>ind_var34_0</th>\n      <th>ind_var34</th>\n      <th>ind_var37_cte</th>\n      <th>ind_var37_0</th>\n      <th>ind_var37</th>\n      <th>ind_var39_0</th>\n      <th>ind_var40_0</th>\n      <th>ind_var40</th>\n      <th>ind_var41_0</th>\n      <th>ind_var41</th>\n      <th>ind_var39</th>\n      <th>ind_var44_0</th>\n      <th>ind_var44</th>\n      <th>ind_var46_0</th>\n      <th>ind_var46</th>\n      <th>num_var1_0</th>\n      <th>num_var1</th>\n      <th>num_var4</th>\n      <th>num_var5_0</th>\n      <th>num_var5</th>\n      <th>num_var6_0</th>\n      <th>num_var6</th>\n      <th>num_var8_0</th>\n      <th>num_var8</th>\n      <th>num_var12_0</th>\n      <th>num_var12</th>\n      <th>num_var13_0</th>\n      <th>num_var13_corto_0</th>\n      <th>num_var13_corto</th>\n      <th>num_var13_largo_0</th>\n      <th>num_var13_largo</th>\n      <th>num_var13_medio_0</th>\n      <th>num_var13_medio</th>\n      <th>num_var13</th>\n      <th>num_var14_0</th>\n      <th>num_var14</th>\n      <th>num_var17_0</th>\n      <th>num_var17</th>\n      <th>num_var18_0</th>\n      <th>num_var18</th>\n      <th>num_var20_0</th>\n      <th>num_var20</th>\n      <th>num_var24_0</th>\n      <th>num_var24</th>\n      <th>num_var26_0</th>\n      <th>num_var26</th>\n      <th>num_var25_0</th>\n      <th>num_var25</th>\n      <th>num_op_var40_hace2</th>\n      <th>num_op_var40_hace3</th>\n      <th>num_op_var40_ult1</th>\n      <th>num_op_var40_ult3</th>\n      <th>num_op_var41_hace2</th>\n      <th>num_op_var41_hace3</th>\n      <th>num_op_var41_ult1</th>\n      <th>num_op_var41_ult3</th>\n      <th>num_op_var39_hace2</th>\n      <th>num_op_var39_hace3</th>\n      <th>num_op_var39_ult1</th>\n      <th>num_op_var39_ult3</th>\n      <th>num_var27_0</th>\n      <th>num_var28_0</th>\n      <th>num_var28</th>\n      <th>num_var27</th>\n      <th>num_var29_0</th>\n      <th>num_var29</th>\n      <th>num_var30_0</th>\n      <th>num_var30</th>\n      <th>num_var31_0</th>\n      <th>num_var31</th>\n      <th>num_var32_0</th>\n      <th>num_var32</th>\n      <th>num_var33_0</th>\n      <th>num_var33</th>\n      <th>num_var34_0</th>\n      <th>num_var34</th>\n      <th>num_var35</th>\n      <th>num_var37_med_ult2</th>\n      <th>num_var37_0</th>\n      <th>num_var37</th>\n      <th>num_var39_0</th>\n      <th>num_var40_0</th>\n      <th>num_var40</th>\n      <th>num_var41_0</th>\n      <th>num_var41</th>\n      <th>num_var39</th>\n      <th>num_var42_0</th>\n      <th>num_var42</th>\n      <th>num_var44_0</th>\n      <th>num_var44</th>\n      <th>num_var46_0</th>\n      <th>num_var46</th>\n      <th>saldo_var1</th>\n      <th>saldo_var5</th>\n      <th>saldo_var6</th>\n      <th>saldo_var8</th>\n      <th>saldo_var12</th>\n      <th>saldo_var13_corto</th>\n      <th>saldo_var13_largo</th>\n      <th>saldo_var13_medio</th>\n      <th>saldo_var13</th>\n      <th>saldo_var14</th>\n      <th>saldo_var17</th>\n      <th>saldo_var18</th>\n      <th>saldo_var20</th>\n      <th>saldo_var24</th>\n      <th>saldo_var26</th>\n      <th>saldo_var25</th>\n      <th>saldo_var28</th>\n      <th>saldo_var27</th>\n      <th>saldo_var29</th>\n      <th>saldo_var30</th>\n      <th>saldo_var31</th>\n      <th>saldo_var32</th>\n      <th>saldo_var33</th>\n      <th>saldo_var34</th>\n      <th>saldo_var37</th>\n      <th>saldo_var40</th>\n      <th>saldo_var41</th>\n      <th>saldo_var42</th>\n      <th>saldo_var44</th>\n      <th>saldo_var46</th>\n      <th>var36</th>\n      <th>delta_imp_amort_var18_1y3</th>\n      <th>delta_imp_amort_var34_1y3</th>\n      <th>delta_imp_aport_var13_1y3</th>\n      <th>delta_imp_aport_var17_1y3</th>\n      <th>delta_imp_aport_var33_1y3</th>\n      <th>delta_imp_compra_var44_1y3</th>\n      <th>delta_imp_reemb_var13_1y3</th>\n      <th>delta_imp_reemb_var17_1y3</th>\n      <th>delta_imp_reemb_var33_1y3</th>\n      <th>delta_imp_trasp_var17_in_1y3</th>\n      <th>delta_imp_trasp_var17_out_1y3</th>\n      <th>delta_imp_trasp_var33_in_1y3</th>\n      <th>delta_imp_trasp_var33_out_1y3</th>\n      <th>delta_imp_venta_var44_1y3</th>\n      <th>delta_num_aport_var13_1y3</th>\n      <th>delta_num_aport_var17_1y3</th>\n      <th>delta_num_aport_var33_1y3</th>\n      <th>delta_num_compra_var44_1y3</th>\n      <th>delta_num_reemb_var13_1y3</th>\n      <th>delta_num_reemb_var17_1y3</th>\n      <th>delta_num_reemb_var33_1y3</th>\n      <th>delta_num_trasp_var17_in_1y3</th>\n      <th>delta_num_trasp_var17_out_1y3</th>\n      <th>delta_num_trasp_var33_in_1y3</th>\n      <th>delta_num_trasp_var33_out_1y3</th>\n      <th>delta_num_venta_var44_1y3</th>\n      <th>imp_amort_var18_hace3</th>\n      <th>imp_amort_var18_ult1</th>\n      <th>imp_amort_var34_hace3</th>\n      <th>imp_amort_var34_ult1</th>\n      <th>imp_aport_var13_hace3</th>\n      <th>imp_aport_var13_ult1</th>\n      <th>imp_aport_var17_hace3</th>\n      <th>imp_aport_var17_ult1</th>\n      <th>imp_aport_var33_hace3</th>\n      <th>imp_aport_var33_ult1</th>\n      <th>imp_var7_emit_ult1</th>\n      <th>imp_var7_recib_ult1</th>\n      <th>imp_compra_var44_hace3</th>\n      <th>imp_compra_var44_ult1</th>\n      <th>imp_reemb_var13_hace3</th>\n      <th>imp_reemb_var13_ult1</th>\n      <th>imp_reemb_var17_hace3</th>\n      <th>imp_reemb_var17_ult1</th>\n      <th>imp_reemb_var33_hace3</th>\n      <th>imp_reemb_var33_ult1</th>\n      <th>imp_var43_emit_ult1</th>\n      <th>imp_trans_var37_ult1</th>\n      <th>imp_trasp_var17_in_hace3</th>\n      <th>imp_trasp_var17_in_ult1</th>\n      <th>imp_trasp_var17_out_hace3</th>\n      <th>imp_trasp_var17_out_ult1</th>\n      <th>imp_trasp_var33_in_hace3</th>\n      <th>imp_trasp_var33_in_ult1</th>\n      <th>imp_trasp_var33_out_hace3</th>\n      <th>imp_trasp_var33_out_ult1</th>\n      <th>imp_venta_var44_hace3</th>\n      <th>imp_venta_var44_ult1</th>\n      <th>ind_var7_emit_ult1</th>\n      <th>ind_var7_recib_ult1</th>\n      <th>ind_var10_ult1</th>\n      <th>ind_var10cte_ult1</th>\n      <th>ind_var9_cte_ult1</th>\n      <th>ind_var9_ult1</th>\n      <th>ind_var43_emit_ult1</th>\n      <th>ind_var43_recib_ult1</th>\n      <th>var21</th>\n      <th>num_var2_0_ult1</th>\n      <th>num_var2_ult1</th>\n      <th>num_aport_var13_hace3</th>\n      <th>num_aport_var13_ult1</th>\n      <th>num_aport_var17_hace3</th>\n      <th>num_aport_var17_ult1</th>\n      <th>num_aport_var33_hace3</th>\n      <th>num_aport_var33_ult1</th>\n      <th>num_var7_emit_ult1</th>\n      <th>num_var7_recib_ult1</th>\n      <th>num_compra_var44_hace3</th>\n      <th>num_compra_var44_ult1</th>\n      <th>num_ent_var16_ult1</th>\n      <th>num_var22_hace2</th>\n      <th>num_var22_hace3</th>\n      <th>num_var22_ult1</th>\n      <th>num_var22_ult3</th>\n      <th>num_med_var22_ult3</th>\n      <th>num_med_var45_ult3</th>\n      <th>num_meses_var5_ult3</th>\n      <th>num_meses_var8_ult3</th>\n      <th>num_meses_var12_ult3</th>\n      <th>num_meses_var13_corto_ult3</th>\n      <th>num_meses_var13_largo_ult3</th>\n      <th>num_meses_var13_medio_ult3</th>\n      <th>num_meses_var17_ult3</th>\n      <th>num_meses_var29_ult3</th>\n      <th>num_meses_var33_ult3</th>\n      <th>num_meses_var39_vig_ult3</th>\n      <th>num_meses_var44_ult3</th>\n      <th>num_op_var39_comer_ult1</th>\n      <th>num_op_var39_comer_ult3</th>\n      <th>num_op_var40_comer_ult1</th>\n      <th>num_op_var40_comer_ult3</th>\n      <th>num_op_var40_efect_ult1</th>\n      <th>num_op_var40_efect_ult3</th>\n      <th>num_op_var41_comer_ult1</th>\n      <th>num_op_var41_comer_ult3</th>\n      <th>num_op_var41_efect_ult1</th>\n      <th>num_op_var41_efect_ult3</th>\n      <th>num_op_var39_efect_ult1</th>\n      <th>num_op_var39_efect_ult3</th>\n      <th>num_reemb_var13_hace3</th>\n      <th>num_reemb_var13_ult1</th>\n      <th>num_reemb_var17_hace3</th>\n      <th>num_reemb_var17_ult1</th>\n      <th>num_reemb_var33_hace3</th>\n      <th>num_reemb_var33_ult1</th>\n      <th>num_sal_var16_ult1</th>\n      <th>num_var43_emit_ult1</th>\n      <th>num_var43_recib_ult1</th>\n      <th>num_trasp_var11_ult1</th>\n      <th>num_trasp_var17_in_hace3</th>\n      <th>num_trasp_var17_in_ult1</th>\n      <th>num_trasp_var17_out_hace3</th>\n      <th>num_trasp_var17_out_ult1</th>\n      <th>num_trasp_var33_in_hace3</th>\n      <th>num_trasp_var33_in_ult1</th>\n      <th>num_trasp_var33_out_hace3</th>\n      <th>num_trasp_var33_out_ult1</th>\n      <th>num_venta_var44_hace3</th>\n      <th>num_venta_var44_ult1</th>\n      <th>num_var45_hace2</th>\n      <th>num_var45_hace3</th>\n      <th>num_var45_ult1</th>\n      <th>num_var45_ult3</th>\n      <th>saldo_var2_ult1</th>\n      <th>saldo_medio_var5_hace2</th>\n      <th>saldo_medio_var5_hace3</th>\n      <th>saldo_medio_var5_ult1</th>\n      <th>saldo_medio_var5_ult3</th>\n      <th>saldo_medio_var8_hace2</th>\n      <th>saldo_medio_var8_hace3</th>\n      <th>saldo_medio_var8_ult1</th>\n      <th>saldo_medio_var8_ult3</th>\n      <th>saldo_medio_var12_hace2</th>\n      <th>saldo_medio_var12_hace3</th>\n      <th>saldo_medio_var12_ult1</th>\n      <th>saldo_medio_var12_ult3</th>\n      <th>saldo_medio_var13_corto_hace2</th>\n      <th>saldo_medio_var13_corto_hace3</th>\n      <th>saldo_medio_var13_corto_ult1</th>\n      <th>saldo_medio_var13_corto_ult3</th>\n      <th>saldo_medio_var13_largo_hace2</th>\n      <th>saldo_medio_var13_largo_hace3</th>\n      <th>saldo_medio_var13_largo_ult1</th>\n      <th>saldo_medio_var13_largo_ult3</th>\n      <th>saldo_medio_var13_medio_hace2</th>\n      <th>saldo_medio_var13_medio_hace3</th>\n      <th>saldo_medio_var13_medio_ult1</th>\n      <th>saldo_medio_var13_medio_ult3</th>\n      <th>saldo_medio_var17_hace2</th>\n      <th>saldo_medio_var17_hace3</th>\n      <th>saldo_medio_var17_ult1</th>\n      <th>saldo_medio_var17_ult3</th>\n      <th>saldo_medio_var29_hace2</th>\n      <th>saldo_medio_var29_hace3</th>\n      <th>saldo_medio_var29_ult1</th>\n      <th>saldo_medio_var29_ult3</th>\n      <th>saldo_medio_var33_hace2</th>\n      <th>saldo_medio_var33_hace3</th>\n      <th>saldo_medio_var33_ult1</th>\n      <th>saldo_medio_var33_ult3</th>\n      <th>saldo_medio_var44_hace2</th>\n      <th>saldo_medio_var44_hace3</th>\n      <th>saldo_medio_var44_ult1</th>\n      <th>saldo_medio_var44_ult3</th>\n      <th>var38</th>\n      <th>TARGET</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1</td>\n      <td>2</td>\n      <td>23</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>99</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.00</td>\n      <td>0.00</td>\n      <td>0.00</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>39205.170000</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>3</td>\n      <td>2</td>\n      <td>34</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>3</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>6</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>300.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>300.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>300.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>-1.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>-1.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>300.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.00</td>\n      <td>88.89</td>\n      <td>0.00</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.00</td>\n      <td>300.0</td>\n      <td>122.22</td>\n      <td>300.0</td>\n      <td>240.75</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>49278.030000</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>4</td>\n      <td>2</td>\n      <td>23</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>3</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>3.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>3.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>3.00</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>99</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3.00</td>\n      <td>0.18</td>\n      <td>3.00</td>\n      <td>2.07</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>67333.770000</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>8</td>\n      <td>2</td>\n      <td>37</td>\n      <td>0.0</td>\n      <td>195.0</td>\n      <td>195.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>195.0</td>\n      <td>195.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>195.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>195.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>3</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>9</td>\n      <td>9</td>\n      <td>0</td>\n      <td>0</td>\n      <td>9</td>\n      <td>9</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>9</td>\n      <td>6</td>\n      <td>6</td>\n      <td>6</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>70.62</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>70.62</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>34.95</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>70.62</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>15</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>9</td>\n      <td>9</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>9</td>\n      <td>9</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>27</td>\n      <td>3</td>\n      <td>18</td>\n      <td>48</td>\n      <td>0</td>\n      <td>186.09</td>\n      <td>0.00</td>\n      <td>91.56</td>\n      <td>138.84</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>64007.970000</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>10</td>\n      <td>2</td>\n      <td>39</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>6</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>6</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>135003.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>135003.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>135003.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>135003.00</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>135003.0</td>\n      <td>270003.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>6</td>\n      <td>9</td>\n      <td>3</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>6</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3.00</td>\n      <td>0.30</td>\n      <td>40501.08</td>\n      <td>13501.47</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>85501.89</td>\n      <td>85501.89</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.00</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>117310.979016</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"markdown","source":"- Uma olhada rapida no dataset nos permite identificar, primeiramente que as variaveis ID e TARGET podem ser retiradas para definir nossa lista de features iniciais. Entretanto, observamos que aparentemente todas as features são numericas. consequentemente não será necesaria a aplicação de métodos de encoding.","metadata":{}},{"cell_type":"code","source":"target = 'TARGET'\nall_features = data.iloc[:,1:-1].columns #retiramos IDs e Target","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:29.848033Z","iopub.execute_input":"2021-09-03T11:34:29.848436Z","iopub.status.idle":"2021-09-03T11:34:29.942832Z","shell.execute_reply.started":"2021-09-03T11:34:29.848401Z","shell.execute_reply":"2021-09-03T11:34:29.941549Z"},"trusted":true},"execution_count":8,"outputs":[]},{"cell_type":"markdown","source":"- Observemos como se encontra distribuida nossa variavel dependente","metadata":{}},{"cell_type":"code","source":"# Histograma da Target\ndata[target].hist()","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:30.719633Z","iopub.execute_input":"2021-09-03T11:34:30.720079Z","iopub.status.idle":"2021-09-03T11:34:31.009860Z","shell.execute_reply.started":"2021-09-03T11:34:30.720045Z","shell.execute_reply":"2021-09-03T11:34:31.008694Z"},"trusted":true},"execution_count":9,"outputs":[{"execution_count":9,"output_type":"execute_result","data":{"text/plain":"<AxesSubplot:>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"see_categories(data, target)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:32.037182Z","iopub.execute_input":"2021-09-03T11:34:32.037740Z","iopub.status.idle":"2021-09-03T11:34:32.069339Z","shell.execute_reply.started":"2021-09-03T11:34:32.037702Z","shell.execute_reply":"2021-09-03T11:34:32.068420Z"},"trusted":true},"execution_count":10,"outputs":[{"execution_count":10,"output_type":"execute_result","data":{"text/plain":"     qtd       perc\n0  73012  96.043147\n1   3008   3.956853","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>qtd</th>\n      <th>perc</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>73012</td>\n      <td>96.043147</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>3008</td>\n      <td>3.956853</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"markdown","source":"<h1> comentarios sobre o desbalanceamento</h1>","metadata":{}},{"cell_type":"markdown","source":"Conforme definido no case, a marcação de 1 indica clientes insatisfeitos e 0 indica clientes satisfeitos <br>\nComo era esperado, a gente como instituição oferece um excelente serviço, por isso a quantidade de clientes satisfeitos é muito maior que os insatisfeitos, o qual faz sentido, e consequentemente o desvalanceamento até que era esperado.<br>\nEntretanto, dado desbalanceamento nos obriga a tomar cuidado com certas ações que tomaremos mais adiante como são: \n- Ao momento de splitar os dados nos datasets train e test é fortemente recomendo utilizar stratificação (stratify) para evitar que exista maior proporção de classes possitivas em um dos datasets.\n- A definição da metrica de avaliação do modelo. Gini, F1-score, ou AUC são boas metricas de classificação onde existe desbalanceamento nos dados. \n- No caso do modelo Random Forest será necessario utilizar o hiperparámetros \"class_weight\" para controlar o desbalanceamento.\n- Recomendado utilizar métodos de UNDERSAMPLING ou OVERSAMPLING.","metadata":{"execution":{"iopub.status.busy":"2021-08-22T20:50:28.54822Z","iopub.execute_input":"2021-08-22T20:50:28.548689Z","iopub.status.idle":"2021-08-22T20:50:28.555552Z","shell.execute_reply.started":"2021-08-22T20:50:28.548649Z","shell.execute_reply":"2021-08-22T20:50:28.554205Z"}}},{"cell_type":"code","source":"X_all = data[all_features] \ny_all = data[target]\n\nX_train, X_test, y_train, y_test = train_test_split(X_all, y_all, stratify = y_all, test_size = 0.20 ,random_state = 1234)\n\nprint(\"Training set has {} samples and {} features.\".format(X_train.shape[0], X_train.shape[1]))\nprint(\"Testing set has {} samples and {} features.\".format(X_test.shape[0], X_test.shape[1]))\nprint(\"Train target {:0.5f}%\".format(y_train.sum()/y_train.shape[0]))\nprint(\"Test target {:0.5f}%\".format(y_test.sum()/y_test.shape[0]))","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:03:23.374598Z","iopub.execute_input":"2021-09-03T19:03:23.375010Z","iopub.status.idle":"2021-09-03T19:03:24.323164Z","shell.execute_reply.started":"2021-09-03T19:03:23.374979Z","shell.execute_reply":"2021-09-03T19:03:24.322146Z"},"trusted":true},"execution_count":144,"outputs":[{"name":"stdout","text":"Training set has 60816 samples and 369 features.\nTesting set has 15204 samples and 369 features.\nTrain target 0.03956%\nTest target 0.03959%\n","output_type":"stream"}]},{"cell_type":"markdown","source":"- Estratificação no target feita com sucesso","metadata":{}},{"cell_type":"markdown","source":"## Verificação de nulos e valores unicos","metadata":{}},{"cell_type":"markdown","source":"- O dataset aparentemente não possui valores nulos, entretanto foi observado a existencia do valor -999999 sendo frequente em algumas variaveis o que levanta a suspeita de que alguem previamente preencheu os valores nulos com o valor -999999 (é uma hipotese).","metadata":{}},{"cell_type":"markdown","source":"- Também obsevamos que existem colunas sem variancia ou variancia muito próxima a zero (aquelas cuja concentração são maioritariamente em uma unica classe).","metadata":{}},{"cell_type":"markdown","source":"- Embora estejamos trabalhando com dados númericos, também é esperado que existam variaveis binarias ou categóricas, aquelas que possuim 2 classes ou poucas classes diferentes.","metadata":{}},{"cell_type":"code","source":"def get_uniques(df):\n    len_array = []\n    concentration_array = []\n    for var in df.columns:\n        current_len  = len(df[var].unique())\n        len_array.append(current_len)\n        concentration  = df[var].value_counts().iloc[0]/df.shape[0]\n        concentration_array.append(concentration)\n\n    missing = pd.DataFrame(np.array(df.isnull().sum()), columns = ['missing'])\n    p_missing = missing / df.shape[0]\n    \n    \n\n    qtd = pd.concat([pd.DataFrame(df.columns), pd.DataFrame(len_array), pd.DataFrame(concentration_array), missing.missing, pd.DataFrame(p_missing)], axis = 1)\n    qtd.columns = ['var', 'qtd_unique_values', 'concentration', 'qtd_missing','perc_missing']\n\n    return qtd\n\nqtd = get_uniques(X_train)\n\nqtd.sort_values('qtd_unique_values')","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:43.165406Z","iopub.execute_input":"2021-09-03T11:34:43.165762Z","iopub.status.idle":"2021-09-03T11:34:43.875258Z","shell.execute_reply.started":"2021-09-03T11:34:43.165731Z","shell.execute_reply":"2021-09-03T11:34:43.874144Z"},"trusted":true},"execution_count":12,"outputs":[{"execution_count":12,"output_type":"execute_result","data":{"text/plain":"                               var  qtd_unique_values  concentration  \\\n315      num_trasp_var17_out_hace3                  1       1.000000   \n131                    num_var27_0                  1       1.000000   \n234          imp_reemb_var13_hace3                  1       1.000000   \n236          imp_reemb_var17_hace3                  1       1.000000   \n327                saldo_var2_ult1                  1       1.000000   \n238          imp_reemb_var33_hace3                  1       1.000000   \n349  saldo_medio_var13_medio_hace3                  1       1.000000   \n261                num_var2_0_ult1                  1       1.000000   \n162                      num_var46                  1       1.000000   \n161                    num_var46_0                  1       1.000000   \n180                    saldo_var27                  1       1.000000   \n155                      num_var41                  1       1.000000   \n244      imp_trasp_var17_out_hace3                  1       1.000000   \n59                       ind_var27                  1       1.000000   \n58                       ind_var28                  1       1.000000   \n248      imp_trasp_var33_out_hace3                  1       1.000000   \n57                     ind_var28_0                  1       1.000000   \n56                     ind_var27_0                  1       1.000000   \n134                      num_var27                  1       1.000000   \n133                      num_var28                  1       1.000000   \n132                    num_var28_0                  1       1.000000   \n80                       ind_var41                  1       1.000000   \n262                  num_var2_ult1                  1       1.000000   \n179                    saldo_var28                  1       1.000000   \n22                        ind_var2                  1       1.000000   \n222          imp_amort_var34_hace3                  1       1.000000   \n305          num_reemb_var17_hace3                  1       1.000000   \n307          num_reemb_var33_hace3                  1       1.000000   \n303          num_reemb_var13_hace3                  1       1.000000   \n319      num_trasp_var33_out_hace3                  1       1.000000   \n192                    saldo_var46                  1       1.000000   \n85                       ind_var46                  1       1.000000   \n84                     ind_var46_0                  1       1.000000   \n220          imp_amort_var18_hace3                  1       1.000000   \n21                      ind_var2_0                  1       1.000000   \n189                    saldo_var41                  1       1.000000   \n91                      num_var6_0                  2       0.999918   \n254                 ind_var10_ult1                  2       0.918919   \n92                        num_var6                  2       0.999967   \n94                        num_var8                  2       0.971323   \n102              num_var13_medio_0                  2       0.999967   \n256              ind_var9_cte_ult1                  2       0.902493   \n103                num_var13_medio                  2       0.999967   \n212      delta_num_reemb_var13_1y3                  2       0.999572   \n257                  ind_var9_ult1                  2       0.913756   \n258            ind_var43_emit_ult1                  2       0.933060   \n109                    num_var18_0                  2       0.999967   \n110                      num_var18                  2       0.999967   \n259           ind_var43_recib_ult1                  2       0.871119   \n111                    num_var20_0                  2       0.996481   \n112                      num_var20                  2       0.997386   \n269             num_var7_emit_ult1                  2       0.999951   \n255              ind_var10cte_ult1                  2       0.907311   \n253            ind_var7_recib_ult1                  2       0.997353   \n316       num_trasp_var17_out_ult1                  2       0.999934   \n135                    num_var29_0                  2       0.999918   \n214      delta_num_reemb_var33_1y3                  2       0.999984   \n206  delta_imp_trasp_var33_out_1y3                  2       0.999984   \n204  delta_imp_trasp_var17_out_1y3                  2       0.999934   \n202      delta_imp_reemb_var33_1y3                  2       0.999984   \n201      delta_imp_reemb_var17_1y3                  2       0.999720   \n200      delta_imp_reemb_var13_1y3                  2       0.999572   \n216  delta_num_trasp_var17_out_1y3                  2       0.999934   \n218  delta_num_trasp_var33_out_1y3                  2       0.999984   \n195      delta_imp_amort_var34_1y3                  2       0.999984   \n194      delta_imp_amort_var18_1y3                  2       0.999967   \n320       num_trasp_var33_out_ult1                  2       0.999984   \n252             ind_var7_emit_ult1                  2       0.999951   \n321          num_venta_var44_hace3                  2       0.999967   \n223           imp_amort_var34_ult1                  2       0.999984   \n239           imp_reemb_var33_ult1                  2       0.999984   \n317       num_trasp_var33_in_hace3                  2       0.999934   \n160                      num_var44                  2       0.998438   \n314        num_trasp_var17_in_ult1                  2       0.999934   \n156                      num_var39                  2       0.996218   \n153                      num_var40                  2       0.996218   \n146                      num_var34                  2       0.999984   \n145                    num_var34_0                  2       0.999984   \n249       imp_trasp_var33_out_ult1                  2       0.999984   \n136                      num_var29                  2       0.999967   \n186                    saldo_var34                  2       0.999984   \n83                       ind_var44                  2       0.998438   \n213      delta_num_reemb_var17_1y3                  2       0.999720   \n81                       ind_var39                  2       0.996218   \n48                     ind_var24_0                  2       0.957215   \n49                       ind_var24                  2       0.961753   \n50                   ind_var25_cte                  2       0.973099   \n51                     ind_var26_0                  2       0.974974   \n52                   ind_var26_cte                  2       0.971932   \n53                       ind_var26                  2       0.974974   \n47                       ind_var20                  2       0.997386   \n54                     ind_var25_0                  2       0.975993   \n60                     ind_var29_0                  2       0.999918   \n61                       ind_var29                  2       0.999967   \n62                     ind_var30_0                  2       0.995791   \n63                       ind_var30                  2       0.733935   \n64                     ind_var31_0                  2       0.995955   \n65                       ind_var31                  2       0.996531   \n55                       ind_var25                  2       0.975993   \n35                 ind_var13_largo                  2       0.990151   \n29                     ind_var12_0                  2       0.932189   \n45                       ind_var19                  2       0.995939   \n34               ind_var13_largo_0                  2       0.990019   \n37                 ind_var13_medio                  2       0.999967   \n38                       ind_var13                  2       0.949109   \n39                     ind_var14_0                  2       0.976437   \n40                       ind_var14                  2       0.994607   \n41                     ind_var17_0                  2       0.998224   \n46                     ind_var20_0                  2       0.996481   \n33                 ind_var13_corto                  2       0.958416   \n42                       ind_var17                  2       0.998569   \n82                     ind_var44_0                  2       0.998241   \n43                     ind_var18_0                  2       0.999967   \n44                       ind_var18                  2       0.999967   \n31                     ind_var13_0                  2       0.947645   \n30                       ind_var12                  2       0.954124   \n32               ind_var13_corto_0                  2       0.956903   \n28                        ind_var8                  2       0.971323   \n66                   ind_var32_cte                  2       0.998767   \n357        saldo_medio_var29_hace3                  2       0.999984   \n79                     ind_var41_0                  2       0.879045   \n78                       ind_var40                  2       0.996218   \n308           num_reemb_var33_ult1                  2       0.999984   \n304           num_reemb_var13_ult1                  2       0.999572   \n77                     ind_var40_0                  2       0.988523   \n76                     ind_var39_0                  2       0.880492   \n75                       ind_var37                  2       0.934853   \n27                      ind_var8_0                  2       0.967114   \n285     num_meses_var13_medio_ult3                  2       0.999967   \n19                      ind_var1_0                  2       0.988490   \n74                     ind_var37_0                  2       0.934853   \n73                   ind_var37_cte                  2       0.927733   \n20                        ind_var1                  2       0.996185   \n36               ind_var13_medio_0                  2       0.999967   \n72                       ind_var34                  2       0.999984   \n23                      ind_var5_0                  2       0.958416   \n24                        ind_var5                  2       0.664957   \n71                     ind_var34_0                  2       0.999984   \n70                       ind_var33                  2       0.999441   \n26                        ind_var6                  2       0.999967   \n67                     ind_var32_0                  2       0.998915   \n25                      ind_var6_0                  2       0.999918   \n69                     ind_var33_0                  2       0.999359   \n68                       ind_var32                  2       0.998915   \n318        num_trasp_var33_in_ult1                  3       0.999934   \n313       num_trasp_var17_in_hace3                  3       0.999967   \n205   delta_imp_trasp_var33_in_1y3                  3       0.999868   \n203   delta_imp_trasp_var17_in_1y3                  3       0.999901   \n181                    saldo_var29                  3       0.999967   \n165                     saldo_var6                  3       0.999967   \n351   saldo_medio_var13_medio_ult3                  3       0.999967   \n221           imp_amort_var18_ult1                  3       0.999967   \n217   delta_num_trasp_var33_in_1y3                  3       0.999868   \n350   saldo_medio_var13_medio_ult1                  3       0.999967   \n359         saldo_medio_var29_ult3                  3       0.999967   \n348  saldo_medio_var13_medio_hace2                  3       0.999967   \n170              saldo_var13_medio                  3       0.999967   \n358         saldo_medio_var29_ult1                  3       0.999967   \n287           num_meses_var29_ult3                  3       0.999934   \n215   delta_num_trasp_var17_in_1y3                  3       0.999901   \n174                    saldo_var18                  3       0.999967   \n106                      num_var14                  3       0.994607   \n87                        num_var1                  3       0.996185   \n99                 num_var13_corto                  3       0.958416   \n98               num_var13_corto_0                  3       0.956903   \n114                      num_var24                  3       0.961753   \n268           num_aport_var33_ult1                  3       0.999901   \n267          num_aport_var33_hace3                  3       0.999737   \n96                       num_var12                  3       0.954124   \n250          imp_venta_var44_hace3                  3       0.999967   \n159                    num_var44_0                  3       0.998241   \n142                      num_var32                  3       0.998915   \n93                      num_var8_0                  3       0.967114   \n144                      num_var33                  3       0.999441   \n86                      num_var1_0                  3       0.988490   \n242       imp_trasp_var17_in_hace3                  3       0.999967   \n152                    num_var40_0                  3       0.988523   \n141                    num_var32_0                  3       0.998915   \n289       num_meses_var39_vig_ult3                  4       0.716637   \n89                      num_var5_0                  4       0.952019   \n286           num_meses_var17_ult3                  4       0.998405   \n283     num_meses_var13_corto_ult3                  4       0.956952   \n284     num_meses_var13_largo_ult3                  4       0.993012   \n290           num_meses_var44_ult3                  4       0.998241   \n219      delta_num_venta_var44_1y3                  4       0.999441   \n356        saldo_medio_var29_hace2                  4       0.999951   \n288           num_meses_var33_ult3                  4       0.999441   \n282           num_meses_var12_ult3                  4       0.953762   \n280            num_meses_var5_ult3                  4       0.564259   \n105                    num_var14_0                  4       0.976437   \n271         num_compra_var44_hace3                  4       0.999556   \n90                        num_var5                  4       0.662572   \n113                    num_var24_0                  4       0.957215   \n230             imp_var7_emit_ult1                  4       0.999951   \n207      delta_imp_venta_var44_1y3                  4       0.999441   \n120             num_op_var40_hace3                  4       0.999918   \n143                    num_var33_0                  4       0.999359   \n210      delta_num_aport_var33_1y3                  4       0.999786   \n281            num_meses_var8_ult3                  4       0.970781   \n306           num_reemb_var17_ult1                  4       0.999720   \n309             num_sal_var16_ult1                  5       0.998717   \n208      delta_num_aport_var13_1y3                  5       0.973198   \n209      delta_num_aport_var17_1y3                  5       0.999277   \n158                      num_var42                  5       0.686185   \n193                          var36                  5       0.394337   \n265          num_aport_var17_hace3                  5       0.999753   \n247        imp_trasp_var33_in_ult1                  5       0.999934   \n246       imp_trasp_var33_in_hace3                  5       0.999934   \n243        imp_trasp_var17_in_ult1                  5       0.999934   \n95                     num_var12_0                  5       0.932189   \n245       imp_trasp_var17_out_ult1                  5       0.999934   \n270            num_var7_recib_ult1                  6       0.997353   \n229           imp_aport_var33_ult1                  6       0.999901   \n263          num_aport_var13_hace3                  6       0.977785   \n157                    num_var42_0                  6       0.924115   \n197      delta_imp_aport_var17_1y3                  6       0.999260   \n118                      num_var25                  7       0.975993   \n264           num_aport_var13_ult1                  7       0.994771   \n296        num_op_var40_efect_ult3                  7       0.999457   \n295        num_op_var40_efect_ult1                  7       0.999556   \n97                     num_var13_0                  7       0.947645   \n100              num_var13_largo_0                  7       0.990019   \n101                num_var13_largo                  7       0.990151   \n117                    num_var25_0                  7       0.975993   \n104                      num_var13                  7       0.949109   \n116                      num_var26                  7       0.974974   \n115                    num_var26_0                  7       0.974974   \n151                    num_var39_0                  8       0.854742   \n154                    num_var41_0                  8       0.860152   \n266           num_aport_var17_ult1                  8       0.999424   \n138                      num_var30                  8       0.677338   \n198      delta_imp_aport_var33_1y3                  8       0.999720   \n88                        num_var4                  8       0.501496   \n322           num_venta_var44_ult1                  8       0.999457   \n272          num_compra_var44_ult1                  9       0.998964   \n140                      num_var31                  9       0.996531   \n108                      num_var17                  9       0.998569   \n211     delta_num_compra_var44_1y3                  9       0.998816   \n137                    num_var30_0                 10       0.895356   \n353        saldo_medio_var17_hace3                 11       0.999836   \n107                    num_var17_0                 11       0.998224   \n139                    num_var31_0                 11       0.995955   \n273             num_ent_var16_ult1                 12       0.951098   \n278             num_med_var22_ult3                 13       0.852062   \n147                      num_var35                 13       0.498586   \n226          imp_aport_var17_hace3                 13       0.999753   \n228          imp_aport_var33_hace3                 14       0.999737   \n312           num_trasp_var11_ult1                 15       0.978345   \n276                 num_var22_ult1                 16       0.894370   \n119             num_op_var40_hace2                 16       0.998783   \n199     delta_imp_compra_var44_1y3                 16       0.998734   \n237           imp_reemb_var17_ult1                 17       0.999720   \n361        saldo_medio_var33_hace3                 17       0.999737   \n7          imp_op_var40_efect_ult1                 18       0.999556   \n275                num_var22_hace3                 19       0.814144   \n196      delta_imp_aport_var13_1y3                 20       0.972968   \n148             num_var37_med_ult2                 20       0.952841   \n8          imp_op_var40_efect_ult3                 21       0.999457   \n128             num_op_var39_hace3                 21       0.986730   \n124             num_op_var41_hace3                 21       0.986813   \n274                num_var22_hace2                 21       0.795366   \n150                      num_var37                 22       0.934853   \n149                    num_var37_0                 22       0.934853   \n235           imp_reemb_var13_ult1                 24       0.999572   \n260                          var21                 24       0.988523   \n310            num_var43_emit_ult1                 25       0.933060   \n299        num_op_var41_efect_ult1                 25       0.911684   \n301        num_op_var39_efect_ult1                 26       0.911569   \n232         imp_compra_var44_hace3                 28       0.999556   \n365        saldo_medio_var44_hace3                 28       0.999556   \n121              num_op_var40_ult1                 29       0.997007   \n293        num_op_var40_comer_ult1                 31       0.996152   \n277                 num_var22_ult3                 31       0.656554   \n360        saldo_medio_var33_hace2                 31       0.999507   \n122              num_op_var40_ult3                 34       0.996777   \n251           imp_venta_var44_ult1                 34       0.999457   \n362         saldo_medio_var33_ult1                 35       0.999441   \n363         saldo_medio_var33_ult3                 35       0.999441   \n227           imp_aport_var17_ult1                 35       0.999424   \n185                    saldo_var33                 35       0.999441   \n311           num_var43_recib_ult1                 35       0.871119   \n300        num_op_var41_efect_ult3                 36       0.885820   \n302        num_op_var39_efect_ult3                 36       0.885672   \n294        num_op_var40_comer_ult3                 43       0.995560   \n127             num_op_var39_hace2                 50       0.889503   \n123             num_op_var41_hace2                 51       0.890341   \n18              imp_sal_var16_ult1                 56       0.998717   \n297        num_op_var41_comer_ult1                 58       0.871251   \n291        num_op_var39_comer_ult1                 61       0.869360   \n184                    saldo_var32                 61       0.998915   \n324                num_var45_hace3                 63       0.664365   \n233          imp_compra_var44_ult1                 64       0.998964   \n125              num_op_var41_ult1                 64       0.846636   \n129              num_op_var39_ult1                 67       0.845107   \n279             num_med_var45_ult3                 68       0.674526   \n352        saldo_medio_var17_hace2                 71       0.998833   \n364        saldo_medio_var44_hace2                 73       0.998800   \n323                num_var45_hace2                 82       0.640424   \n298        num_op_var41_comer_ult3                 84       0.848313   \n325                 num_var45_ult1                 88       0.714615   \n173                    saldo_var17                 88       0.998569   \n292        num_op_var39_comer_ult3                 88       0.846422   \n126              num_op_var41_ult3                 93       0.819735   \n355         saldo_medio_var17_ult3                 94       0.998471   \n354         saldo_medio_var17_ult1                 94       0.998471   \n191                    saldo_var44                 95       0.998438   \n130              num_op_var39_ult3                 97       0.818272   \n1                            var15                 99       0.264651   \n367         saldo_medio_var44_ult3                105       0.998273   \n366         saldo_medio_var44_ult1                105       0.998273   \n175                    saldo_var20                142       0.997386   \n231            imp_var7_recib_ult1                148       0.997353   \n225           imp_aport_var13_ult1                159       0.994771   \n326                 num_var45_ult3                159       0.502532   \n9                imp_op_var40_ult1                180       0.997007   \n346   saldo_medio_var13_largo_ult1                183       0.993094   \n0                             var3                189       0.975911   \n169              saldo_var13_largo                190       0.990151   \n345  saldo_medio_var13_largo_hace3                209       0.996202   \n183                    saldo_var31                211       0.996531   \n188                    saldo_var40                229       0.996218   \n163                     saldo_var1                231       0.996185   \n5          imp_op_var40_comer_ult1                234       0.996152   \n344  saldo_medio_var13_largo_hace2                242       0.993653   \n6          imp_op_var40_comer_ult3                269       0.995560   \n172                    saldo_var14                285       0.994607   \n12         imp_op_var41_efect_ult1                297       0.911684   \n15         imp_op_var39_efect_ult1                301       0.911569   \n333         saldo_medio_var8_hace3                343       0.993900   \n224          imp_aport_var13_hace3                359       0.977785   \n13         imp_op_var41_efect_ult3                403       0.885820   \n347   saldo_medio_var13_largo_ult3                405       0.993094   \n16         imp_op_var39_efect_ult3                409       0.885672   \n2               imp_ent_var16_ult1                522       0.951098   \n168              saldo_var13_corto                628       0.958416   \n171                    saldo_var13                733       0.949109   \n342   saldo_medio_var13_corto_ult1                789       0.958070   \n341  saldo_medio_var13_corto_hace3                802       0.982225   \n337        saldo_medio_var12_hace3                941       0.983031   \n332         saldo_medio_var8_hace2               1067       0.981715   \n178                    saldo_var25               1246       0.975993   \n177                    saldo_var26               1302       0.974974   \n340  saldo_medio_var13_corto_hace2               1358       0.960800   \n166                     saldo_var8               1610       0.971323   \n334          saldo_medio_var8_ult1               1704       0.970994   \n335          saldo_medio_var8_ult3               1743       0.970994   \n240            imp_var43_emit_ult1               1954       0.933060   \n343   saldo_medio_var13_corto_ult3               2125       0.958070   \n176                    saldo_var24               2140       0.961753   \n336        saldo_medio_var12_hace2               2293       0.960339   \n167                    saldo_var12               2508       0.954124   \n338         saldo_medio_var12_ult1               2756       0.953844   \n339         saldo_medio_var12_ult3               2785       0.953844   \n241           imp_trans_var37_ult1               3197       0.871119   \n187                    saldo_var37               3303       0.934853   \n10         imp_op_var41_comer_ult1               6104       0.871251   \n3          imp_op_var39_comer_ult1               6207       0.869360   \n14               imp_op_var41_ult1               6557       0.846636   \n329         saldo_medio_var5_hace3               6576       0.388894   \n17               imp_op_var39_ult1               6651       0.845107   \n11         imp_op_var41_comer_ult3               7303       0.848313   \n4          imp_op_var39_comer_ult3               7411       0.846422   \n164                     saldo_var5               9682       0.335043   \n328         saldo_medio_var5_hace2              12171       0.303818   \n330          saldo_medio_var5_ult1              12307       0.323024   \n190                    saldo_var42              13017       0.285155   \n182                    saldo_var30              14023       0.266065   \n331          saldo_medio_var5_ult3              14539       0.323024   \n368                          var38              46666       0.194850   \n\n     qtd_missing  perc_missing  \n315            0           0.0  \n131            0           0.0  \n234            0           0.0  \n236            0           0.0  \n327            0           0.0  \n238            0           0.0  \n349            0           0.0  \n261            0           0.0  \n162            0           0.0  \n161            0           0.0  \n180            0           0.0  \n155            0           0.0  \n244            0           0.0  \n59             0           0.0  \n58             0           0.0  \n248            0           0.0  \n57             0           0.0  \n56             0           0.0  \n134            0           0.0  \n133            0           0.0  \n132            0           0.0  \n80             0           0.0  \n262            0           0.0  \n179            0           0.0  \n22             0           0.0  \n222            0           0.0  \n305            0           0.0  \n307            0           0.0  \n303            0           0.0  \n319            0           0.0  \n192            0           0.0  \n85             0           0.0  \n84             0           0.0  \n220            0           0.0  \n21             0           0.0  \n189            0           0.0  \n91             0           0.0  \n254            0           0.0  \n92             0           0.0  \n94             0           0.0  \n102            0           0.0  \n256            0           0.0  \n103            0           0.0  \n212            0           0.0  \n257            0           0.0  \n258            0           0.0  \n109            0           0.0  \n110            0           0.0  \n259            0           0.0  \n111            0           0.0  \n112            0           0.0  \n269            0           0.0  \n255            0           0.0  \n253            0           0.0  \n316            0           0.0  \n135            0           0.0  \n214            0           0.0  \n206            0           0.0  \n204            0           0.0  \n202            0           0.0  \n201            0           0.0  \n200            0           0.0  \n216            0           0.0  \n218            0           0.0  \n195            0           0.0  \n194            0           0.0  \n320            0           0.0  \n252            0           0.0  \n321            0           0.0  \n223            0           0.0  \n239            0           0.0  \n317            0           0.0  \n160            0           0.0  \n314            0           0.0  \n156            0           0.0  \n153            0           0.0  \n146            0           0.0  \n145            0           0.0  \n249            0           0.0  \n136            0           0.0  \n186            0           0.0  \n83             0           0.0  \n213            0           0.0  \n81             0           0.0  \n48             0           0.0  \n49             0           0.0  \n50             0           0.0  \n51             0           0.0  \n52             0           0.0  \n53             0           0.0  \n47             0           0.0  \n54             0           0.0  \n60             0           0.0  \n61             0           0.0  \n62             0           0.0  \n63             0           0.0  \n64             0           0.0  \n65             0           0.0  \n55             0           0.0  \n35             0           0.0  \n29             0           0.0  \n45             0           0.0  \n34             0           0.0  \n37             0           0.0  \n38             0           0.0  \n39             0           0.0  \n40             0           0.0  \n41             0           0.0  \n46             0           0.0  \n33             0           0.0  \n42             0           0.0  \n82             0           0.0  \n43             0           0.0  \n44             0           0.0  \n31             0           0.0  \n30             0           0.0  \n32             0           0.0  \n28             0           0.0  \n66             0           0.0  \n357            0           0.0  \n79             0           0.0  \n78             0           0.0  \n308            0           0.0  \n304            0           0.0  \n77             0           0.0  \n76             0           0.0  \n75             0           0.0  \n27             0           0.0  \n285            0           0.0  \n19             0           0.0  \n74             0           0.0  \n73             0           0.0  \n20             0           0.0  \n36             0           0.0  \n72             0           0.0  \n23             0           0.0  \n24             0           0.0  \n71             0           0.0  \n70             0           0.0  \n26             0           0.0  \n67             0           0.0  \n25             0           0.0  \n69             0           0.0  \n68             0           0.0  \n318            0           0.0  \n313            0           0.0  \n205            0           0.0  \n203            0           0.0  \n181            0           0.0  \n165            0           0.0  \n351            0           0.0  \n221            0           0.0  \n217            0           0.0  \n350            0           0.0  \n359            0           0.0  \n348            0           0.0  \n170            0           0.0  \n358            0           0.0  \n287            0           0.0  \n215            0           0.0  \n174            0           0.0  \n106            0           0.0  \n87             0           0.0  \n99             0           0.0  \n98             0           0.0  \n114            0           0.0  \n268            0           0.0  \n267            0           0.0  \n96             0           0.0  \n250            0           0.0  \n159            0           0.0  \n142            0           0.0  \n93             0           0.0  \n144            0           0.0  \n86             0           0.0  \n242            0           0.0  \n152            0           0.0  \n141            0           0.0  \n289            0           0.0  \n89             0           0.0  \n286            0           0.0  \n283            0           0.0  \n284            0           0.0  \n290            0           0.0  \n219            0           0.0  \n356            0           0.0  \n288            0           0.0  \n282            0           0.0  \n280            0           0.0  \n105            0           0.0  \n271            0           0.0  \n90             0           0.0  \n113            0           0.0  \n230            0           0.0  \n207            0           0.0  \n120            0           0.0  \n143            0           0.0  \n210            0           0.0  \n281            0           0.0  \n306            0           0.0  \n309            0           0.0  \n208            0           0.0  \n209            0           0.0  \n158            0           0.0  \n193            0           0.0  \n265            0           0.0  \n247            0           0.0  \n246            0           0.0  \n243            0           0.0  \n95             0           0.0  \n245            0           0.0  \n270            0           0.0  \n229            0           0.0  \n263            0           0.0  \n157            0           0.0  \n197            0           0.0  \n118            0           0.0  \n264            0           0.0  \n296            0           0.0  \n295            0           0.0  \n97             0           0.0  \n100            0           0.0  \n101            0           0.0  \n117            0           0.0  \n104            0           0.0  \n116            0           0.0  \n115            0           0.0  \n151            0           0.0  \n154            0           0.0  \n266            0           0.0  \n138            0           0.0  \n198            0           0.0  \n88             0           0.0  \n322            0           0.0  \n272            0           0.0  \n140            0           0.0  \n108            0           0.0  \n211            0           0.0  \n137            0           0.0  \n353            0           0.0  \n107            0           0.0  \n139            0           0.0  \n273            0           0.0  \n278            0           0.0  \n147            0           0.0  \n226            0           0.0  \n228            0           0.0  \n312            0           0.0  \n276            0           0.0  \n119            0           0.0  \n199            0           0.0  \n237            0           0.0  \n361            0           0.0  \n7              0           0.0  \n275            0           0.0  \n196            0           0.0  \n148            0           0.0  \n8              0           0.0  \n128            0           0.0  \n124            0           0.0  \n274            0           0.0  \n150            0           0.0  \n149            0           0.0  \n235            0           0.0  \n260            0           0.0  \n310            0           0.0  \n299            0           0.0  \n301            0           0.0  \n232            0           0.0  \n365            0           0.0  \n121            0           0.0  \n293            0           0.0  \n277            0           0.0  \n360            0           0.0  \n122            0           0.0  \n251            0           0.0  \n362            0           0.0  \n363            0           0.0  \n227            0           0.0  \n185            0           0.0  \n311            0           0.0  \n300            0           0.0  \n302            0           0.0  \n294            0           0.0  \n127            0           0.0  \n123            0           0.0  \n18             0           0.0  \n297            0           0.0  \n291            0           0.0  \n184            0           0.0  \n324            0           0.0  \n233            0           0.0  \n125            0           0.0  \n129            0           0.0  \n279            0           0.0  \n352            0           0.0  \n364            0           0.0  \n323            0           0.0  \n298            0           0.0  \n325            0           0.0  \n173            0           0.0  \n292            0           0.0  \n126            0           0.0  \n355            0           0.0  \n354            0           0.0  \n191            0           0.0  \n130            0           0.0  \n1              0           0.0  \n367            0           0.0  \n366            0           0.0  \n175            0           0.0  \n231            0           0.0  \n225            0           0.0  \n326            0           0.0  \n9              0           0.0  \n346            0           0.0  \n0              0           0.0  \n169            0           0.0  \n345            0           0.0  \n183            0           0.0  \n188            0           0.0  \n163            0           0.0  \n5              0           0.0  \n344            0           0.0  \n6              0           0.0  \n172            0           0.0  \n12             0           0.0  \n15             0           0.0  \n333            0           0.0  \n224            0           0.0  \n13             0           0.0  \n347            0           0.0  \n16             0           0.0  \n2              0           0.0  \n168            0           0.0  \n171            0           0.0  \n342            0           0.0  \n341            0           0.0  \n337            0           0.0  \n332            0           0.0  \n178            0           0.0  \n177            0           0.0  \n340            0           0.0  \n166            0           0.0  \n334            0           0.0  \n335            0           0.0  \n240            0           0.0  \n343            0           0.0  \n176            0           0.0  \n336            0           0.0  \n167            0           0.0  \n338            0           0.0  \n339            0           0.0  \n241            0           0.0  \n187            0           0.0  \n10             0           0.0  \n3              0           0.0  \n14             0           0.0  \n329            0           0.0  \n17             0           0.0  \n11             0           0.0  \n4              0           0.0  \n164            0           0.0  \n328            0           0.0  \n330            0           0.0  \n190            0           0.0  \n182            0           0.0  \n331            0           0.0  \n368            0           0.0  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>var</th>\n      <th>qtd_unique_values</th>\n      <th>concentration</th>\n      <th>qtd_missing</th>\n      <th>perc_missing</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>315</th>\n      <td>num_trasp_var17_out_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>131</th>\n      <td>num_var27_0</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>234</th>\n      <td>imp_reemb_var13_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>236</th>\n      <td>imp_reemb_var17_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>327</th>\n      <td>saldo_var2_ult1</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>238</th>\n      <td>imp_reemb_var33_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>349</th>\n      <td>saldo_medio_var13_medio_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>261</th>\n      <td>num_var2_0_ult1</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>162</th>\n      <td>num_var46</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>161</th>\n      <td>num_var46_0</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>180</th>\n      <td>saldo_var27</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>155</th>\n      <td>num_var41</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>244</th>\n      <td>imp_trasp_var17_out_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>59</th>\n      <td>ind_var27</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>58</th>\n      <td>ind_var28</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>248</th>\n      <td>imp_trasp_var33_out_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>57</th>\n      <td>ind_var28_0</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>56</th>\n      <td>ind_var27_0</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>134</th>\n      <td>num_var27</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>133</th>\n      <td>num_var28</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>132</th>\n      <td>num_var28_0</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>80</th>\n      <td>ind_var41</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>262</th>\n      <td>num_var2_ult1</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>179</th>\n      <td>saldo_var28</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>22</th>\n      <td>ind_var2</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>222</th>\n      <td>imp_amort_var34_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>305</th>\n      <td>num_reemb_var17_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>307</th>\n      <td>num_reemb_var33_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>303</th>\n      <td>num_reemb_var13_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>319</th>\n      <td>num_trasp_var33_out_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>192</th>\n      <td>saldo_var46</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>85</th>\n      <td>ind_var46</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>84</th>\n      <td>ind_var46_0</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>220</th>\n      <td>imp_amort_var18_hace3</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>21</th>\n      <td>ind_var2_0</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>189</th>\n      <td>saldo_var41</td>\n      <td>1</td>\n      <td>1.000000</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>91</th>\n      <td>num_var6_0</td>\n      <td>2</td>\n      <td>0.999918</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>254</th>\n      <td>ind_var10_ult1</td>\n      <td>2</td>\n      <td>0.918919</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>92</th>\n      <td>num_var6</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>94</th>\n      <td>num_var8</td>\n      <td>2</td>\n      <td>0.971323</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>102</th>\n      <td>num_var13_medio_0</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>256</th>\n      <td>ind_var9_cte_ult1</td>\n      <td>2</td>\n      <td>0.902493</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>103</th>\n      <td>num_var13_medio</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>212</th>\n      <td>delta_num_reemb_var13_1y3</td>\n      <td>2</td>\n      <td>0.999572</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>257</th>\n      <td>ind_var9_ult1</td>\n      <td>2</td>\n      <td>0.913756</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>258</th>\n      <td>ind_var43_emit_ult1</td>\n      <td>2</td>\n      <td>0.933060</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>109</th>\n      <td>num_var18_0</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>110</th>\n      <td>num_var18</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>259</th>\n      <td>ind_var43_recib_ult1</td>\n      <td>2</td>\n      <td>0.871119</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>111</th>\n      <td>num_var20_0</td>\n      <td>2</td>\n      <td>0.996481</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>112</th>\n      <td>num_var20</td>\n      <td>2</td>\n      <td>0.997386</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>269</th>\n      <td>num_var7_emit_ult1</td>\n      <td>2</td>\n      <td>0.999951</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>255</th>\n      <td>ind_var10cte_ult1</td>\n      <td>2</td>\n      <td>0.907311</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>253</th>\n      <td>ind_var7_recib_ult1</td>\n      <td>2</td>\n      <td>0.997353</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>316</th>\n      <td>num_trasp_var17_out_ult1</td>\n      <td>2</td>\n      <td>0.999934</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>135</th>\n      <td>num_var29_0</td>\n      <td>2</td>\n      <td>0.999918</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>214</th>\n      <td>delta_num_reemb_var33_1y3</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>206</th>\n      <td>delta_imp_trasp_var33_out_1y3</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>204</th>\n      <td>delta_imp_trasp_var17_out_1y3</td>\n      <td>2</td>\n      <td>0.999934</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>202</th>\n      <td>delta_imp_reemb_var33_1y3</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>201</th>\n      <td>delta_imp_reemb_var17_1y3</td>\n      <td>2</td>\n      <td>0.999720</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>200</th>\n      <td>delta_imp_reemb_var13_1y3</td>\n      <td>2</td>\n      <td>0.999572</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>216</th>\n      <td>delta_num_trasp_var17_out_1y3</td>\n      <td>2</td>\n      <td>0.999934</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>218</th>\n      <td>delta_num_trasp_var33_out_1y3</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>195</th>\n      <td>delta_imp_amort_var34_1y3</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>194</th>\n      <td>delta_imp_amort_var18_1y3</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>320</th>\n      <td>num_trasp_var33_out_ult1</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>252</th>\n      <td>ind_var7_emit_ult1</td>\n      <td>2</td>\n      <td>0.999951</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>321</th>\n      <td>num_venta_var44_hace3</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>223</th>\n      <td>imp_amort_var34_ult1</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>239</th>\n      <td>imp_reemb_var33_ult1</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>317</th>\n      <td>num_trasp_var33_in_hace3</td>\n      <td>2</td>\n      <td>0.999934</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>160</th>\n      <td>num_var44</td>\n      <td>2</td>\n      <td>0.998438</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>314</th>\n      <td>num_trasp_var17_in_ult1</td>\n      <td>2</td>\n      <td>0.999934</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>156</th>\n      <td>num_var39</td>\n      <td>2</td>\n      <td>0.996218</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>153</th>\n      <td>num_var40</td>\n      <td>2</td>\n      <td>0.996218</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>146</th>\n      <td>num_var34</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>145</th>\n      <td>num_var34_0</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>249</th>\n      <td>imp_trasp_var33_out_ult1</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>136</th>\n      <td>num_var29</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>186</th>\n      <td>saldo_var34</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>83</th>\n      <td>ind_var44</td>\n      <td>2</td>\n      <td>0.998438</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>213</th>\n      <td>delta_num_reemb_var17_1y3</td>\n      <td>2</td>\n      <td>0.999720</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>81</th>\n      <td>ind_var39</td>\n      <td>2</td>\n      <td>0.996218</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>48</th>\n      <td>ind_var24_0</td>\n      <td>2</td>\n      <td>0.957215</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>49</th>\n      <td>ind_var24</td>\n      <td>2</td>\n      <td>0.961753</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>50</th>\n      <td>ind_var25_cte</td>\n      <td>2</td>\n      <td>0.973099</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>51</th>\n      <td>ind_var26_0</td>\n      <td>2</td>\n      <td>0.974974</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>52</th>\n      <td>ind_var26_cte</td>\n      <td>2</td>\n      <td>0.971932</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>53</th>\n      <td>ind_var26</td>\n      <td>2</td>\n      <td>0.974974</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>47</th>\n      <td>ind_var20</td>\n      <td>2</td>\n      <td>0.997386</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>54</th>\n      <td>ind_var25_0</td>\n      <td>2</td>\n      <td>0.975993</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>60</th>\n      <td>ind_var29_0</td>\n      <td>2</td>\n      <td>0.999918</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>61</th>\n      <td>ind_var29</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>62</th>\n      <td>ind_var30_0</td>\n      <td>2</td>\n      <td>0.995791</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>63</th>\n      <td>ind_var30</td>\n      <td>2</td>\n      <td>0.733935</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>64</th>\n      <td>ind_var31_0</td>\n      <td>2</td>\n      <td>0.995955</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>65</th>\n      <td>ind_var31</td>\n      <td>2</td>\n      <td>0.996531</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>55</th>\n      <td>ind_var25</td>\n      <td>2</td>\n      <td>0.975993</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>35</th>\n      <td>ind_var13_largo</td>\n      <td>2</td>\n      <td>0.990151</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>29</th>\n      <td>ind_var12_0</td>\n      <td>2</td>\n      <td>0.932189</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>45</th>\n      <td>ind_var19</td>\n      <td>2</td>\n      <td>0.995939</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>34</th>\n      <td>ind_var13_largo_0</td>\n      <td>2</td>\n      <td>0.990019</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>37</th>\n      <td>ind_var13_medio</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>38</th>\n      <td>ind_var13</td>\n      <td>2</td>\n      <td>0.949109</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>39</th>\n      <td>ind_var14_0</td>\n      <td>2</td>\n      <td>0.976437</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>40</th>\n      <td>ind_var14</td>\n      <td>2</td>\n      <td>0.994607</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>41</th>\n      <td>ind_var17_0</td>\n      <td>2</td>\n      <td>0.998224</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>46</th>\n      <td>ind_var20_0</td>\n      <td>2</td>\n      <td>0.996481</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>33</th>\n      <td>ind_var13_corto</td>\n      <td>2</td>\n      <td>0.958416</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>42</th>\n      <td>ind_var17</td>\n      <td>2</td>\n      <td>0.998569</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>82</th>\n      <td>ind_var44_0</td>\n      <td>2</td>\n      <td>0.998241</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>43</th>\n      <td>ind_var18_0</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>44</th>\n      <td>ind_var18</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>31</th>\n      <td>ind_var13_0</td>\n      <td>2</td>\n      <td>0.947645</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>30</th>\n      <td>ind_var12</td>\n      <td>2</td>\n      <td>0.954124</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>32</th>\n      <td>ind_var13_corto_0</td>\n      <td>2</td>\n      <td>0.956903</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>28</th>\n      <td>ind_var8</td>\n      <td>2</td>\n      <td>0.971323</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>66</th>\n      <td>ind_var32_cte</td>\n      <td>2</td>\n      <td>0.998767</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>357</th>\n      <td>saldo_medio_var29_hace3</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>79</th>\n      <td>ind_var41_0</td>\n      <td>2</td>\n      <td>0.879045</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>78</th>\n      <td>ind_var40</td>\n      <td>2</td>\n      <td>0.996218</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>308</th>\n      <td>num_reemb_var33_ult1</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>304</th>\n      <td>num_reemb_var13_ult1</td>\n      <td>2</td>\n      <td>0.999572</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>77</th>\n      <td>ind_var40_0</td>\n      <td>2</td>\n      <td>0.988523</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>76</th>\n      <td>ind_var39_0</td>\n      <td>2</td>\n      <td>0.880492</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>75</th>\n      <td>ind_var37</td>\n      <td>2</td>\n      <td>0.934853</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>27</th>\n      <td>ind_var8_0</td>\n      <td>2</td>\n      <td>0.967114</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>285</th>\n      <td>num_meses_var13_medio_ult3</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>19</th>\n      <td>ind_var1_0</td>\n      <td>2</td>\n      <td>0.988490</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>74</th>\n      <td>ind_var37_0</td>\n      <td>2</td>\n      <td>0.934853</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>73</th>\n      <td>ind_var37_cte</td>\n      <td>2</td>\n      <td>0.927733</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>20</th>\n      <td>ind_var1</td>\n      <td>2</td>\n      <td>0.996185</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>36</th>\n      <td>ind_var13_medio_0</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>72</th>\n      <td>ind_var34</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>23</th>\n      <td>ind_var5_0</td>\n      <td>2</td>\n      <td>0.958416</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>24</th>\n      <td>ind_var5</td>\n      <td>2</td>\n      <td>0.664957</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>71</th>\n      <td>ind_var34_0</td>\n      <td>2</td>\n      <td>0.999984</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>70</th>\n      <td>ind_var33</td>\n      <td>2</td>\n      <td>0.999441</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>26</th>\n      <td>ind_var6</td>\n      <td>2</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>67</th>\n      <td>ind_var32_0</td>\n      <td>2</td>\n      <td>0.998915</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>25</th>\n      <td>ind_var6_0</td>\n      <td>2</td>\n      <td>0.999918</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>69</th>\n      <td>ind_var33_0</td>\n      <td>2</td>\n      <td>0.999359</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>68</th>\n      <td>ind_var32</td>\n      <td>2</td>\n      <td>0.998915</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>318</th>\n      <td>num_trasp_var33_in_ult1</td>\n      <td>3</td>\n      <td>0.999934</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>313</th>\n      <td>num_trasp_var17_in_hace3</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>205</th>\n      <td>delta_imp_trasp_var33_in_1y3</td>\n      <td>3</td>\n      <td>0.999868</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>203</th>\n      <td>delta_imp_trasp_var17_in_1y3</td>\n      <td>3</td>\n      <td>0.999901</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>181</th>\n      <td>saldo_var29</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>165</th>\n      <td>saldo_var6</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>351</th>\n      <td>saldo_medio_var13_medio_ult3</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>221</th>\n      <td>imp_amort_var18_ult1</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>217</th>\n      <td>delta_num_trasp_var33_in_1y3</td>\n      <td>3</td>\n      <td>0.999868</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>350</th>\n      <td>saldo_medio_var13_medio_ult1</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>359</th>\n      <td>saldo_medio_var29_ult3</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>348</th>\n      <td>saldo_medio_var13_medio_hace2</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>170</th>\n      <td>saldo_var13_medio</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>358</th>\n      <td>saldo_medio_var29_ult1</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>287</th>\n      <td>num_meses_var29_ult3</td>\n      <td>3</td>\n      <td>0.999934</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>215</th>\n      <td>delta_num_trasp_var17_in_1y3</td>\n      <td>3</td>\n      <td>0.999901</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>174</th>\n      <td>saldo_var18</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>106</th>\n      <td>num_var14</td>\n      <td>3</td>\n      <td>0.994607</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>87</th>\n      <td>num_var1</td>\n      <td>3</td>\n      <td>0.996185</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>99</th>\n      <td>num_var13_corto</td>\n      <td>3</td>\n      <td>0.958416</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>98</th>\n      <td>num_var13_corto_0</td>\n      <td>3</td>\n      <td>0.956903</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>114</th>\n      <td>num_var24</td>\n      <td>3</td>\n      <td>0.961753</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>268</th>\n      <td>num_aport_var33_ult1</td>\n      <td>3</td>\n      <td>0.999901</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>267</th>\n      <td>num_aport_var33_hace3</td>\n      <td>3</td>\n      <td>0.999737</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>96</th>\n      <td>num_var12</td>\n      <td>3</td>\n      <td>0.954124</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>250</th>\n      <td>imp_venta_var44_hace3</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>159</th>\n      <td>num_var44_0</td>\n      <td>3</td>\n      <td>0.998241</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>142</th>\n      <td>num_var32</td>\n      <td>3</td>\n      <td>0.998915</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>93</th>\n      <td>num_var8_0</td>\n      <td>3</td>\n      <td>0.967114</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>144</th>\n      <td>num_var33</td>\n      <td>3</td>\n      <td>0.999441</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>86</th>\n      <td>num_var1_0</td>\n      <td>3</td>\n      <td>0.988490</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>242</th>\n      <td>imp_trasp_var17_in_hace3</td>\n      <td>3</td>\n      <td>0.999967</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>152</th>\n      <td>num_var40_0</td>\n      <td>3</td>\n      <td>0.988523</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>141</th>\n      <td>num_var32_0</td>\n      <td>3</td>\n      <td>0.998915</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>289</th>\n      <td>num_meses_var39_vig_ult3</td>\n      <td>4</td>\n      <td>0.716637</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>89</th>\n      <td>num_var5_0</td>\n      <td>4</td>\n      <td>0.952019</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>286</th>\n      <td>num_meses_var17_ult3</td>\n      <td>4</td>\n      <td>0.998405</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>283</th>\n      <td>num_meses_var13_corto_ult3</td>\n      <td>4</td>\n      <td>0.956952</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>284</th>\n      <td>num_meses_var13_largo_ult3</td>\n      <td>4</td>\n      <td>0.993012</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>290</th>\n      <td>num_meses_var44_ult3</td>\n      <td>4</td>\n      <td>0.998241</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>219</th>\n      <td>delta_num_venta_var44_1y3</td>\n      <td>4</td>\n      <td>0.999441</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>356</th>\n      <td>saldo_medio_var29_hace2</td>\n      <td>4</td>\n      <td>0.999951</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>288</th>\n      <td>num_meses_var33_ult3</td>\n      <td>4</td>\n      <td>0.999441</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>282</th>\n      <td>num_meses_var12_ult3</td>\n      <td>4</td>\n      <td>0.953762</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>280</th>\n      <td>num_meses_var5_ult3</td>\n      <td>4</td>\n      <td>0.564259</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>105</th>\n      <td>num_var14_0</td>\n      <td>4</td>\n      <td>0.976437</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>271</th>\n      <td>num_compra_var44_hace3</td>\n      <td>4</td>\n      <td>0.999556</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>90</th>\n      <td>num_var5</td>\n      <td>4</td>\n      <td>0.662572</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>113</th>\n      <td>num_var24_0</td>\n      <td>4</td>\n      <td>0.957215</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>230</th>\n      <td>imp_var7_emit_ult1</td>\n      <td>4</td>\n      <td>0.999951</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>207</th>\n      <td>delta_imp_venta_var44_1y3</td>\n      <td>4</td>\n      <td>0.999441</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>120</th>\n      <td>num_op_var40_hace3</td>\n      <td>4</td>\n      <td>0.999918</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>143</th>\n      <td>num_var33_0</td>\n      <td>4</td>\n      <td>0.999359</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>210</th>\n      <td>delta_num_aport_var33_1y3</td>\n      <td>4</td>\n      <td>0.999786</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>281</th>\n      <td>num_meses_var8_ult3</td>\n      <td>4</td>\n      <td>0.970781</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>306</th>\n      <td>num_reemb_var17_ult1</td>\n      <td>4</td>\n      <td>0.999720</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>309</th>\n      <td>num_sal_var16_ult1</td>\n      <td>5</td>\n      <td>0.998717</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>208</th>\n      <td>delta_num_aport_var13_1y3</td>\n      <td>5</td>\n      <td>0.973198</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>209</th>\n      <td>delta_num_aport_var17_1y3</td>\n      <td>5</td>\n      <td>0.999277</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>158</th>\n      <td>num_var42</td>\n      <td>5</td>\n      <td>0.686185</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>193</th>\n      <td>var36</td>\n      <td>5</td>\n      <td>0.394337</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>265</th>\n      <td>num_aport_var17_hace3</td>\n      <td>5</td>\n      <td>0.999753</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>247</th>\n      <td>imp_trasp_var33_in_ult1</td>\n      <td>5</td>\n      <td>0.999934</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>246</th>\n      <td>imp_trasp_var33_in_hace3</td>\n      <td>5</td>\n      <td>0.999934</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>243</th>\n      <td>imp_trasp_var17_in_ult1</td>\n      <td>5</td>\n      <td>0.999934</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>95</th>\n      <td>num_var12_0</td>\n      <td>5</td>\n      <td>0.932189</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>245</th>\n      <td>imp_trasp_var17_out_ult1</td>\n      <td>5</td>\n      <td>0.999934</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>270</th>\n      <td>num_var7_recib_ult1</td>\n      <td>6</td>\n      <td>0.997353</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>229</th>\n      <td>imp_aport_var33_ult1</td>\n      <td>6</td>\n      <td>0.999901</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>263</th>\n      <td>num_aport_var13_hace3</td>\n      <td>6</td>\n      <td>0.977785</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>157</th>\n      <td>num_var42_0</td>\n      <td>6</td>\n      <td>0.924115</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>197</th>\n      <td>delta_imp_aport_var17_1y3</td>\n      <td>6</td>\n      <td>0.999260</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>118</th>\n      <td>num_var25</td>\n      <td>7</td>\n      <td>0.975993</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>264</th>\n      <td>num_aport_var13_ult1</td>\n      <td>7</td>\n      <td>0.994771</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>296</th>\n      <td>num_op_var40_efect_ult3</td>\n      <td>7</td>\n      <td>0.999457</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>295</th>\n      <td>num_op_var40_efect_ult1</td>\n      <td>7</td>\n      <td>0.999556</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>97</th>\n      <td>num_var13_0</td>\n      <td>7</td>\n      <td>0.947645</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>100</th>\n      <td>num_var13_largo_0</td>\n      <td>7</td>\n      <td>0.990019</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>101</th>\n      <td>num_var13_largo</td>\n      <td>7</td>\n      <td>0.990151</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>117</th>\n      <td>num_var25_0</td>\n      <td>7</td>\n      <td>0.975993</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>104</th>\n      <td>num_var13</td>\n      <td>7</td>\n      <td>0.949109</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>116</th>\n      <td>num_var26</td>\n      <td>7</td>\n      <td>0.974974</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>115</th>\n      <td>num_var26_0</td>\n      <td>7</td>\n      <td>0.974974</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>151</th>\n      <td>num_var39_0</td>\n      <td>8</td>\n      <td>0.854742</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>154</th>\n      <td>num_var41_0</td>\n      <td>8</td>\n      <td>0.860152</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>266</th>\n      <td>num_aport_var17_ult1</td>\n      <td>8</td>\n      <td>0.999424</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>138</th>\n      <td>num_var30</td>\n      <td>8</td>\n      <td>0.677338</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>198</th>\n      <td>delta_imp_aport_var33_1y3</td>\n      <td>8</td>\n      <td>0.999720</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>88</th>\n      <td>num_var4</td>\n      <td>8</td>\n      <td>0.501496</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>322</th>\n      <td>num_venta_var44_ult1</td>\n      <td>8</td>\n      <td>0.999457</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>272</th>\n      <td>num_compra_var44_ult1</td>\n      <td>9</td>\n      <td>0.998964</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>140</th>\n      <td>num_var31</td>\n      <td>9</td>\n      <td>0.996531</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>108</th>\n      <td>num_var17</td>\n      <td>9</td>\n      <td>0.998569</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>211</th>\n      <td>delta_num_compra_var44_1y3</td>\n      <td>9</td>\n      <td>0.998816</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>137</th>\n      <td>num_var30_0</td>\n      <td>10</td>\n      <td>0.895356</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>353</th>\n      <td>saldo_medio_var17_hace3</td>\n      <td>11</td>\n      <td>0.999836</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>107</th>\n      <td>num_var17_0</td>\n      <td>11</td>\n      <td>0.998224</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>139</th>\n      <td>num_var31_0</td>\n      <td>11</td>\n      <td>0.995955</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>273</th>\n      <td>num_ent_var16_ult1</td>\n      <td>12</td>\n      <td>0.951098</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>278</th>\n      <td>num_med_var22_ult3</td>\n      <td>13</td>\n      <td>0.852062</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>147</th>\n      <td>num_var35</td>\n      <td>13</td>\n      <td>0.498586</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>226</th>\n      <td>imp_aport_var17_hace3</td>\n      <td>13</td>\n      <td>0.999753</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>228</th>\n      <td>imp_aport_var33_hace3</td>\n      <td>14</td>\n      <td>0.999737</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>312</th>\n      <td>num_trasp_var11_ult1</td>\n      <td>15</td>\n      <td>0.978345</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>276</th>\n      <td>num_var22_ult1</td>\n      <td>16</td>\n      <td>0.894370</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>119</th>\n      <td>num_op_var40_hace2</td>\n      <td>16</td>\n      <td>0.998783</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>199</th>\n      <td>delta_imp_compra_var44_1y3</td>\n      <td>16</td>\n      <td>0.998734</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>237</th>\n      <td>imp_reemb_var17_ult1</td>\n      <td>17</td>\n      <td>0.999720</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>361</th>\n      <td>saldo_medio_var33_hace3</td>\n      <td>17</td>\n      <td>0.999737</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>7</th>\n      <td>imp_op_var40_efect_ult1</td>\n      <td>18</td>\n      <td>0.999556</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>275</th>\n      <td>num_var22_hace3</td>\n      <td>19</td>\n      <td>0.814144</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>196</th>\n      <td>delta_imp_aport_var13_1y3</td>\n      <td>20</td>\n      <td>0.972968</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>148</th>\n      <td>num_var37_med_ult2</td>\n      <td>20</td>\n      <td>0.952841</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>8</th>\n      <td>imp_op_var40_efect_ult3</td>\n      <td>21</td>\n      <td>0.999457</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>128</th>\n      <td>num_op_var39_hace3</td>\n      <td>21</td>\n      <td>0.986730</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>124</th>\n      <td>num_op_var41_hace3</td>\n      <td>21</td>\n      <td>0.986813</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>274</th>\n      <td>num_var22_hace2</td>\n      <td>21</td>\n      <td>0.795366</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>150</th>\n      <td>num_var37</td>\n      <td>22</td>\n      <td>0.934853</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>149</th>\n      <td>num_var37_0</td>\n      <td>22</td>\n      <td>0.934853</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>235</th>\n      <td>imp_reemb_var13_ult1</td>\n      <td>24</td>\n      <td>0.999572</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>260</th>\n      <td>var21</td>\n      <td>24</td>\n      <td>0.988523</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>310</th>\n      <td>num_var43_emit_ult1</td>\n      <td>25</td>\n      <td>0.933060</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>299</th>\n      <td>num_op_var41_efect_ult1</td>\n      <td>25</td>\n      <td>0.911684</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>301</th>\n      <td>num_op_var39_efect_ult1</td>\n      <td>26</td>\n      <td>0.911569</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>232</th>\n      <td>imp_compra_var44_hace3</td>\n      <td>28</td>\n      <td>0.999556</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>365</th>\n      <td>saldo_medio_var44_hace3</td>\n      <td>28</td>\n      <td>0.999556</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>121</th>\n      <td>num_op_var40_ult1</td>\n      <td>29</td>\n      <td>0.997007</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>293</th>\n      <td>num_op_var40_comer_ult1</td>\n      <td>31</td>\n      <td>0.996152</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>277</th>\n      <td>num_var22_ult3</td>\n      <td>31</td>\n      <td>0.656554</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>360</th>\n      <td>saldo_medio_var33_hace2</td>\n      <td>31</td>\n      <td>0.999507</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>122</th>\n      <td>num_op_var40_ult3</td>\n      <td>34</td>\n      <td>0.996777</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>251</th>\n      <td>imp_venta_var44_ult1</td>\n      <td>34</td>\n      <td>0.999457</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>362</th>\n      <td>saldo_medio_var33_ult1</td>\n      <td>35</td>\n      <td>0.999441</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>363</th>\n      <td>saldo_medio_var33_ult3</td>\n      <td>35</td>\n      <td>0.999441</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>227</th>\n      <td>imp_aport_var17_ult1</td>\n      <td>35</td>\n      <td>0.999424</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>185</th>\n      <td>saldo_var33</td>\n      <td>35</td>\n      <td>0.999441</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>311</th>\n      <td>num_var43_recib_ult1</td>\n      <td>35</td>\n      <td>0.871119</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>300</th>\n      <td>num_op_var41_efect_ult3</td>\n      <td>36</td>\n      <td>0.885820</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>302</th>\n      <td>num_op_var39_efect_ult3</td>\n      <td>36</td>\n      <td>0.885672</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>294</th>\n      <td>num_op_var40_comer_ult3</td>\n      <td>43</td>\n      <td>0.995560</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>127</th>\n      <td>num_op_var39_hace2</td>\n      <td>50</td>\n      <td>0.889503</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>123</th>\n      <td>num_op_var41_hace2</td>\n      <td>51</td>\n      <td>0.890341</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>18</th>\n      <td>imp_sal_var16_ult1</td>\n      <td>56</td>\n      <td>0.998717</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>297</th>\n      <td>num_op_var41_comer_ult1</td>\n      <td>58</td>\n      <td>0.871251</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>291</th>\n      <td>num_op_var39_comer_ult1</td>\n      <td>61</td>\n      <td>0.869360</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>184</th>\n      <td>saldo_var32</td>\n      <td>61</td>\n      <td>0.998915</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>324</th>\n      <td>num_var45_hace3</td>\n      <td>63</td>\n      <td>0.664365</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>233</th>\n      <td>imp_compra_var44_ult1</td>\n      <td>64</td>\n      <td>0.998964</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>125</th>\n      <td>num_op_var41_ult1</td>\n      <td>64</td>\n      <td>0.846636</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>129</th>\n      <td>num_op_var39_ult1</td>\n      <td>67</td>\n      <td>0.845107</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>279</th>\n      <td>num_med_var45_ult3</td>\n      <td>68</td>\n      <td>0.674526</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>352</th>\n      <td>saldo_medio_var17_hace2</td>\n      <td>71</td>\n      <td>0.998833</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>364</th>\n      <td>saldo_medio_var44_hace2</td>\n      <td>73</td>\n      <td>0.998800</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>323</th>\n      <td>num_var45_hace2</td>\n      <td>82</td>\n      <td>0.640424</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>298</th>\n      <td>num_op_var41_comer_ult3</td>\n      <td>84</td>\n      <td>0.848313</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>325</th>\n      <td>num_var45_ult1</td>\n      <td>88</td>\n      <td>0.714615</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>173</th>\n      <td>saldo_var17</td>\n      <td>88</td>\n      <td>0.998569</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>292</th>\n      <td>num_op_var39_comer_ult3</td>\n      <td>88</td>\n      <td>0.846422</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>126</th>\n      <td>num_op_var41_ult3</td>\n      <td>93</td>\n      <td>0.819735</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>355</th>\n      <td>saldo_medio_var17_ult3</td>\n      <td>94</td>\n      <td>0.998471</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>354</th>\n      <td>saldo_medio_var17_ult1</td>\n      <td>94</td>\n      <td>0.998471</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>191</th>\n      <td>saldo_var44</td>\n      <td>95</td>\n      <td>0.998438</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>130</th>\n      <td>num_op_var39_ult3</td>\n      <td>97</td>\n      <td>0.818272</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>var15</td>\n      <td>99</td>\n      <td>0.264651</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>367</th>\n      <td>saldo_medio_var44_ult3</td>\n      <td>105</td>\n      <td>0.998273</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>366</th>\n      <td>saldo_medio_var44_ult1</td>\n      <td>105</td>\n      <td>0.998273</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>175</th>\n      <td>saldo_var20</td>\n      <td>142</td>\n      <td>0.997386</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>231</th>\n      <td>imp_var7_recib_ult1</td>\n      <td>148</td>\n      <td>0.997353</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>225</th>\n      <td>imp_aport_var13_ult1</td>\n      <td>159</td>\n      <td>0.994771</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>326</th>\n      <td>num_var45_ult3</td>\n      <td>159</td>\n      <td>0.502532</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>9</th>\n      <td>imp_op_var40_ult1</td>\n      <td>180</td>\n      <td>0.997007</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>346</th>\n      <td>saldo_medio_var13_largo_ult1</td>\n      <td>183</td>\n      <td>0.993094</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>0</th>\n      <td>var3</td>\n      <td>189</td>\n      <td>0.975911</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>169</th>\n      <td>saldo_var13_largo</td>\n      <td>190</td>\n      <td>0.990151</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>345</th>\n      <td>saldo_medio_var13_largo_hace3</td>\n      <td>209</td>\n      <td>0.996202</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>183</th>\n      <td>saldo_var31</td>\n      <td>211</td>\n      <td>0.996531</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>188</th>\n      <td>saldo_var40</td>\n      <td>229</td>\n      <td>0.996218</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>163</th>\n      <td>saldo_var1</td>\n      <td>231</td>\n      <td>0.996185</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>imp_op_var40_comer_ult1</td>\n      <td>234</td>\n      <td>0.996152</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>344</th>\n      <td>saldo_medio_var13_largo_hace2</td>\n      <td>242</td>\n      <td>0.993653</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>6</th>\n      <td>imp_op_var40_comer_ult3</td>\n      <td>269</td>\n      <td>0.995560</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>172</th>\n      <td>saldo_var14</td>\n      <td>285</td>\n      <td>0.994607</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>12</th>\n      <td>imp_op_var41_efect_ult1</td>\n      <td>297</td>\n      <td>0.911684</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>15</th>\n      <td>imp_op_var39_efect_ult1</td>\n      <td>301</td>\n      <td>0.911569</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>333</th>\n      <td>saldo_medio_var8_hace3</td>\n      <td>343</td>\n      <td>0.993900</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>224</th>\n      <td>imp_aport_var13_hace3</td>\n      <td>359</td>\n      <td>0.977785</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>13</th>\n      <td>imp_op_var41_efect_ult3</td>\n      <td>403</td>\n      <td>0.885820</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>347</th>\n      <td>saldo_medio_var13_largo_ult3</td>\n      <td>405</td>\n      <td>0.993094</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>16</th>\n      <td>imp_op_var39_efect_ult3</td>\n      <td>409</td>\n      <td>0.885672</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>imp_ent_var16_ult1</td>\n      <td>522</td>\n      <td>0.951098</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>168</th>\n      <td>saldo_var13_corto</td>\n      <td>628</td>\n      <td>0.958416</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>171</th>\n      <td>saldo_var13</td>\n      <td>733</td>\n      <td>0.949109</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>342</th>\n      <td>saldo_medio_var13_corto_ult1</td>\n      <td>789</td>\n      <td>0.958070</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>341</th>\n      <td>saldo_medio_var13_corto_hace3</td>\n      <td>802</td>\n      <td>0.982225</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>337</th>\n      <td>saldo_medio_var12_hace3</td>\n      <td>941</td>\n      <td>0.983031</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>332</th>\n      <td>saldo_medio_var8_hace2</td>\n      <td>1067</td>\n      <td>0.981715</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>178</th>\n      <td>saldo_var25</td>\n      <td>1246</td>\n      <td>0.975993</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>177</th>\n      <td>saldo_var26</td>\n      <td>1302</td>\n      <td>0.974974</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>340</th>\n      <td>saldo_medio_var13_corto_hace2</td>\n      <td>1358</td>\n      <td>0.960800</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>166</th>\n      <td>saldo_var8</td>\n      <td>1610</td>\n      <td>0.971323</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>334</th>\n      <td>saldo_medio_var8_ult1</td>\n      <td>1704</td>\n      <td>0.970994</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>335</th>\n      <td>saldo_medio_var8_ult3</td>\n      <td>1743</td>\n      <td>0.970994</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>240</th>\n      <td>imp_var43_emit_ult1</td>\n      <td>1954</td>\n      <td>0.933060</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>343</th>\n      <td>saldo_medio_var13_corto_ult3</td>\n      <td>2125</td>\n      <td>0.958070</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>176</th>\n      <td>saldo_var24</td>\n      <td>2140</td>\n      <td>0.961753</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>336</th>\n      <td>saldo_medio_var12_hace2</td>\n      <td>2293</td>\n      <td>0.960339</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>167</th>\n      <td>saldo_var12</td>\n      <td>2508</td>\n      <td>0.954124</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>338</th>\n      <td>saldo_medio_var12_ult1</td>\n      <td>2756</td>\n      <td>0.953844</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>339</th>\n      <td>saldo_medio_var12_ult3</td>\n      <td>2785</td>\n      <td>0.953844</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>241</th>\n      <td>imp_trans_var37_ult1</td>\n      <td>3197</td>\n      <td>0.871119</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>187</th>\n      <td>saldo_var37</td>\n      <td>3303</td>\n      <td>0.934853</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>10</th>\n      <td>imp_op_var41_comer_ult1</td>\n      <td>6104</td>\n      <td>0.871251</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>imp_op_var39_comer_ult1</td>\n      <td>6207</td>\n      <td>0.869360</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>14</th>\n      <td>imp_op_var41_ult1</td>\n      <td>6557</td>\n      <td>0.846636</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>329</th>\n      <td>saldo_medio_var5_hace3</td>\n      <td>6576</td>\n      <td>0.388894</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>17</th>\n      <td>imp_op_var39_ult1</td>\n      <td>6651</td>\n      <td>0.845107</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>11</th>\n      <td>imp_op_var41_comer_ult3</td>\n      <td>7303</td>\n      <td>0.848313</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>imp_op_var39_comer_ult3</td>\n      <td>7411</td>\n      <td>0.846422</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>164</th>\n      <td>saldo_var5</td>\n      <td>9682</td>\n      <td>0.335043</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>328</th>\n      <td>saldo_medio_var5_hace2</td>\n      <td>12171</td>\n      <td>0.303818</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>330</th>\n      <td>saldo_medio_var5_ult1</td>\n      <td>12307</td>\n      <td>0.323024</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>190</th>\n      <td>saldo_var42</td>\n      <td>13017</td>\n      <td>0.285155</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>182</th>\n      <td>saldo_var30</td>\n      <td>14023</td>\n      <td>0.266065</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>331</th>\n      <td>saldo_medio_var5_ult3</td>\n      <td>14539</td>\n      <td>0.323024</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>368</th>\n      <td>var38</td>\n      <td>46666</td>\n      <td>0.194850</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"markdown","source":"Por enquanto só retiraremos as colunas sem variancia e colunas duplicadas.","metadata":{}},{"cell_type":"code","source":"zero_variance = list(qtd[qtd['qtd_unique_values'] == 1]['var'])\nlen(zero_variance)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:47.516037Z","iopub.execute_input":"2021-09-03T11:34:47.516562Z","iopub.status.idle":"2021-09-03T11:34:47.523536Z","shell.execute_reply.started":"2021-09-03T11:34:47.516531Z","shell.execute_reply":"2021-09-03T11:34:47.522820Z"},"trusted":true},"execution_count":13,"outputs":[{"execution_count":13,"output_type":"execute_result","data":{"text/plain":"36"},"metadata":{}}]},{"cell_type":"code","source":"selected_features = list(set(X_train.columns) - set(zero_variance))\nlen(selected_features)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:48.005684Z","iopub.execute_input":"2021-09-03T11:34:48.006184Z","iopub.status.idle":"2021-09-03T11:34:48.011697Z","shell.execute_reply.started":"2021-09-03T11:34:48.006152Z","shell.execute_reply":"2021-09-03T11:34:48.010855Z"},"trusted":true},"execution_count":14,"outputs":[{"execution_count":14,"output_type":"execute_result","data":{"text/plain":"333"},"metadata":{}}]},{"cell_type":"code","source":"aux = []\naux2 = {}\nc = selected_features\nfor i in range(len(c)-1):\n    if c[i] not in list(dict.fromkeys(sum(aux2.values(), []))):\n        v = X_train[c[i]].values\n        aux2[c[i]] = []\n        for j in range(i+1,len(c)):\n            if np.array_equal(v,X_train[c[j]].values):\n                aux.append([c[i], c[j]])\n                aux2[c[i]].append(c[j])\n\nfor i in aux2:\n    if len(aux2[i]) >= 1:\n        print(i, '\\t', aux2[i])\n        \nduplicate_vars = list(dict.fromkeys(sum(aux2.values(), [])))\nlen(duplicate_vars)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:49.941397Z","iopub.execute_input":"2021-09-03T11:34:49.941727Z","iopub.status.idle":"2021-09-03T11:34:55.133778Z","shell.execute_reply.started":"2021-09-03T11:34:49.941699Z","shell.execute_reply":"2021-09-03T11:34:55.132766Z"},"trusted":true},"execution_count":15,"outputs":[{"name":"stdout","text":"num_var39 \t ['num_var40']\ndelta_num_trasp_var33_in_1y3 \t ['delta_imp_trasp_var33_in_1y3']\nind_var13_medio_0 \t ['ind_var13_medio']\nnum_var29_0 \t ['num_var6_0']\ndelta_imp_reemb_var13_1y3 \t ['delta_num_reemb_var13_1y3']\nind_var32_0 \t ['ind_var32']\nind_var18 \t ['ind_var18_0']\ndelta_imp_reemb_var17_1y3 \t ['delta_num_reemb_var17_1y3']\nnum_var34_0 \t ['num_var34']\nind_var34 \t ['ind_var34_0']\ndelta_num_trasp_var33_out_1y3 \t ['delta_imp_trasp_var33_out_1y3']\nnum_var25_0 \t ['num_var25']\ndelta_imp_trasp_var17_out_1y3 \t ['delta_num_trasp_var17_out_1y3']\ndelta_num_reemb_var33_1y3 \t ['delta_imp_reemb_var33_1y3']\ndelta_imp_trasp_var17_in_1y3 \t ['delta_num_trasp_var17_in_1y3']\nsaldo_medio_var13_medio_ult1 \t ['saldo_var13_medio']\nsaldo_var6 \t ['saldo_var29']\nind_var26_0 \t ['ind_var26']\nind_var29 \t ['ind_var6']\nind_var39 \t ['ind_var40']\nind_var37_0 \t ['ind_var37']\nnum_var13_medio_0 \t ['num_var13_medio']\nind_var6_0 \t ['ind_var29_0']\nnum_var37_0 \t ['num_var37']\nnum_var26_0 \t ['num_var26']\nnum_var29 \t ['num_var6']\nnum_var32_0 \t ['num_var32']\nind_var25_0 \t ['ind_var25']\nnum_var18_0 \t ['num_var18']\n","output_type":"stream"},{"execution_count":15,"output_type":"execute_result","data":{"text/plain":"29"},"metadata":{}}]},{"cell_type":"code","source":"selected_features = list(set(X_train.columns) \n                         - set(zero_variance) \n                         - set(duplicate_vars)\n                        )\nlen(selected_features)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:55.136979Z","iopub.execute_input":"2021-09-03T11:34:55.137276Z","iopub.status.idle":"2021-09-03T11:34:55.143969Z","shell.execute_reply.started":"2021-09-03T11:34:55.137247Z","shell.execute_reply":"2021-09-03T11:34:55.142953Z"},"trusted":true},"execution_count":16,"outputs":[{"execution_count":16,"output_type":"execute_result","data":{"text/plain":"304"},"metadata":{}}]},{"cell_type":"markdown","source":"# Tarefa 1 (Maximizar o Lucro)\n\n- Agora definiremos nossa função de custo/ganho a qual procuraremos minimizar/maximizar. (Neste caso iremos a maximizar o lucro conforme regras de negocio)\n- Regras de negocio:\n - O custo de preparar e executar uma ação de retenção é um valor fixo de R$ 10 por cliente.\n - Um verdadeiro positivo é um cliente que estava insatisfeito e foi alvo de uma ação de retenção. O benefício neste caso é o lucro da ação (R\\\\$ 100) menos os custos relacionados à ação de retenção (R\\\\$ 10).\n - Um verdadeiro negativo é um cliente insatisfeito e que não é alvo de nenhuma ação. O benefício neste caso é zero, isto é, nenhum custo, mas nenhum lucro.\n - Um falso positivo ocorre quando classificamos um cliente como insatisfeito, mas ela não se comporta como tal, neste caso o valor da ação de retenção será perdido.\n - Um falso negativo ocorre quando um cliente é previsto como satisfeito, mas na verdade ele estava insatisfeito. Neste caso, nenhum dinheiro foi gasto e nada foi ganho.","metadata":{"execution":{"iopub.status.busy":"2021-08-22T20:44:48.352977Z","iopub.execute_input":"2021-08-22T20:44:48.35341Z","iopub.status.idle":"2021-08-22T20:44:48.360132Z","shell.execute_reply.started":"2021-08-22T20:44:48.353372Z","shell.execute_reply":"2021-08-22T20:44:48.358589Z"}}},{"cell_type":"markdown","source":"Nossa função de lucro vai depender unicamente dos classificados como falsos possitivos (FP) e verdadeiros possitivos (TP). Para obter os FP e VP utilizamos a função confusion_matrix().","metadata":{}},{"cell_type":"code","source":"# Funcao proposta por CaioMar \ndef max_profit_score(y_true, y_pred):\n    \"\"\"\n    Função que avalia o modelo computando o lucro esperado\n    da ação de retenção dos clientes insatisfeitos.\n    \"\"\"\n    conf_matrix = confusion_matrix(y_true, y_pred)\n    \n    FP = conf_matrix[0][1] # Falsos positivos\n    TP = conf_matrix[1][1] # Verdadeiros positivos\n    \n    lucro = 90*TP - 10*FP # lucro da acao de retencao (os \n                          # verdadeiros e falsos negativos\n                          # não contribuem em\n                          # nada no lucro)\n    \n    return lucro\n\n\ndef show_performance_2(clf, X, y_true):\n    y_pred = clf.predict(X)\n    lucro_max_possible = max_profit_score(y_true,y_true)\n    return 100*max_profit_score(y_true, y_pred)/lucro_max_possible\n\n\ndef show_performance(clf, X, y_true):\n    y_pred = clf.predict(X)\n    lucro_max_possible = max_profit_score(y_true,y_true)\n    if str(type(clf)) == \"<class 'lightgbm.basic.Booster'>\":\n        y_pred = (clf.predict(X_test) > 0.5)*1\n    print('Matriz de confusão:')\n    print(confusion_matrix(y_true, y_pred))\n    print('\\nLucro obtido com a ação (em Reais): R$ {} '.format(max_profit_score(y_true, y_pred)))\n    print('Lucro maximo possivel: R$ {} '.format(lucro_max_possible))\n    print('Porcentagem do lucro maximo: {:0.2f}%'.format(100*max_profit_score(y_true, y_pred)/lucro_max_possible))\n    print('')\n\n\nmax_profit_scorer = make_scorer(max_profit_score, greater_is_better=True)\n\nprint('Lucro máximo possivel na partição de treino: {}'.format(max_profit_score(data[target], data[target])))\nprint('Lucro máximo possivel na partição de teste: {}'.format(max_profit_score(y_test, y_test)))","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:34:58.802953Z","iopub.execute_input":"2021-09-03T11:34:58.803309Z","iopub.status.idle":"2021-09-03T11:34:58.945913Z","shell.execute_reply.started":"2021-09-03T11:34:58.803265Z","shell.execute_reply":"2021-09-03T11:34:58.944854Z"},"trusted":true},"execution_count":17,"outputs":[{"name":"stdout","text":"Lucro máximo possivel na partição de treino: 270720\nLucro máximo possivel na partição de teste: 54180\n","output_type":"stream"}]},{"cell_type":"markdown","source":"Procedemos a treinar nosso modelo, e como ponto de partida utilizamos um RandomForestClassifier. por que Random Forest? particularmente sempre recomendamos iniciar com RandomForest dado que ele oferece boa performance com um baixo custo de treinamento, é altamente paralelizavel, performa bem inclusive quando os dados possuim pouco tratamento e é robusto a outliers, trabalha bem com variaveis contínuas e categoricas, entre outras vantagens. ","metadata":{}},{"cell_type":"markdown","source":"Um passo importante e cruzial na hora de instanciar qualquer modelo é a escolha dos hiperparametros, isso pode ser um motivo que diferencia um modelo bom e um ruim. Um metodo que funciona muito bem que é muito conhecido é GridSearchCV que seria o equivalente a aplicar \"forza bruta\". Entretanto, neste momento optaremos por utilizar scikit-Optimize (SKOPT). SKOPT providencia uma estrutura para tuning de hiperparametros pela abordagem de otimização bayesiana. Alguns trabalhos cientificos provam que otimização bayesiana é mais eficiente que grid search atingindo modelos de igual ou melhor performance.","metadata":{}},{"cell_type":"markdown","source":"Para entender o conceito e o efeito dos hiperparametros do RandomForest é recomendada a leitura da referencia https://www.analyticsvidhya.com/blog/2020/03/beginners-guide-random-forest-hyperparameter-tuning/.","metadata":{}},{"cell_type":"markdown","source":"Definiremos alguns parametros, que pasaremos ao modelo para ele entender melhor o que estamos tentando otimizar. \nNeste é utilizada a função make_score para indicar ao modelo que temos que maximizar nossa função de lucro (greater_is_better=True) conforme indicado pelo CaioMar.","metadata":{}},{"cell_type":"code","source":"%%time \n\nimport skopt\nfrom sklearn.model_selection import cross_val_score\n\ndef tune_rf(params):\n    \"\"\"Function to be passed as scikit-optimize minimizer/maximizer input\n    \n    Parameters:\n    Tuples with information about the range that the optimizer should use for that parameter, \n    as well as the behaviour that it should follow in that range.\n    \n    Returns:\n    float: the metric that should be minimized. If the objective is maximization, then the negative \n    of the desired metric must be returned. In this case, the negative AUC average generated by CV is returned.\n    \"\"\"\n    \n    \n    #Hyperparameters to be optimized\n    print(params)\n    max_depth = params[0] \n    n_estimators = params[1] \n    class_weight = params[2]\n    min_samples_split = params[3]\n    max_features = params[4]\n\n        \n    \n    #Model to be optimized\n    model = RandomForestClassifier(\n                            max_depth = max_depth,\n                            n_estimators = n_estimators,\n                            class_weight = {0:1,1:class_weight},\n                            min_samples_split = min_samples_split,\n                            max_features = max_features,\n                            n_jobs=-1,\n                            random_state=1234\n            )\n    \n\n    \n\n    \n    #Cross-Validation in order to avoid overfitting\n    lucro_perc = cross_val_score(model, X_train[selected_features], y_train, cv = 5, scoring = max_profit_scorer)\n    \n    print(lucro_perc.mean())\n    # as the function is minimization (forest_minimize), we need to use the negative of the desired metric (AUC)\n    return -lucro_perc.mean()\n\n\nspace = [(18, 19), # max_depth\n          (46, 47), # n_estimators\n          (10,15), # class_weight \n          (int(X_train.shape[0]*0.005), int(X_train.shape[0]*0.01)), # min_samples_split \n          ('auto','log2', 40)] # max_features\n\n\n\nresult = skopt.forest_minimize(tune_rf, space, random_state=42, n_random_starts=20, n_calls=50, verbose=1)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T11:36:29.472421Z","iopub.execute_input":"2021-09-03T11:36:29.472906Z","iopub.status.idle":"2021-09-03T11:50:39.595756Z","shell.execute_reply.started":"2021-09-03T11:36:29.472877Z","shell.execute_reply":"2021-09-03T11:50:39.594978Z"},"trusted":true},"execution_count":21,"outputs":[{"name":"stdout","text":"Iteration No: 1 started. Evaluating function at random point.\n[19, 46, 12, 375, 'log2']\n10332.0\nIteration No: 1 ended. Evaluation done at random point.\nTime taken: 10.1835\nFunction value obtained: -10332.0000\nCurrent minimum: -10332.0000\nIteration No: 2 started. Evaluating function at random point.\n[18, 47, 12, 518, 'auto']\n10934.0\nIteration No: 2 ended. Evaluation done at random point.\nTime taken: 10.9355\nFunction value obtained: -10934.0000\nCurrent minimum: -10934.0000\nIteration No: 3 started. Evaluating function at random point.\n[19, 46, 13, 455, 'log2']\n10544.0\nIteration No: 3 ended. Evaluation done at random point.\nTime taken: 7.8197\nFunction value obtained: -10544.0000\nCurrent minimum: -10934.0000\nIteration No: 4 started. Evaluating function at random point.\n[18, 47, 13, 597, 'auto']\n10818.0\nIteration No: 4 ended. Evaluation done at random point.\nTime taken: 11.1036\nFunction value obtained: -10818.0000\nCurrent minimum: -10934.0000\nIteration No: 5 started. Evaluating function at random point.\n[19, 46, 10, 325, 'auto']\n10894.0\nIteration No: 5 ended. Evaluation done at random point.\nTime taken: 11.5469\nFunction value obtained: -10894.0000\nCurrent minimum: -10934.0000\nIteration No: 6 started. Evaluating function at random point.\n[18, 46, 12, 362, 'log2']\n10608.0\nIteration No: 6 ended. Evaluation done at random point.\nTime taken: 7.7939\nFunction value obtained: -10608.0000\nCurrent minimum: -10934.0000\nIteration No: 7 started. Evaluating function at random point.\n[19, 47, 13, 574, 'log2']\n10304.0\nIteration No: 7 ended. Evaluation done at random point.\nTime taken: 8.0526\nFunction value obtained: -10304.0000\nCurrent minimum: -10934.0000\nIteration No: 8 started. Evaluating function at random point.\n[18, 47, 12, 358, 40]\n11242.0\nIteration No: 8 ended. Evaluation done at random point.\nTime taken: 20.3427\nFunction value obtained: -11242.0000\nCurrent minimum: -11242.0000\nIteration No: 9 started. Evaluating function at random point.\n[18, 46, 14, 438, 'auto']\n11036.0\nIteration No: 9 ended. Evaluation done at random point.\nTime taken: 11.4515\nFunction value obtained: -11036.0000\nCurrent minimum: -11242.0000\nIteration No: 10 started. Evaluating function at random point.\n[18, 47, 13, 392, 40]\n11506.0\nIteration No: 10 ended. Evaluation done at random point.\nTime taken: 20.0070\nFunction value obtained: -11506.0000\nCurrent minimum: -11506.0000\nIteration No: 11 started. Evaluating function at random point.\n[19, 47, 10, 356, 'auto']\n10814.0\nIteration No: 11 ended. Evaluation done at random point.\nTime taken: 12.0308\nFunction value obtained: -10814.0000\nCurrent minimum: -11506.0000\nIteration No: 12 started. Evaluating function at random point.\n[19, 46, 13, 567, 'auto']\n10888.0\nIteration No: 12 ended. Evaluation done at random point.\nTime taken: 11.7038\nFunction value obtained: -10888.0000\nCurrent minimum: -11506.0000\nIteration No: 13 started. Evaluating function at random point.\n[19, 46, 13, 353, 'log2']\n10512.0\nIteration No: 13 ended. Evaluation done at random point.\nTime taken: 8.1778\nFunction value obtained: -10512.0000\nCurrent minimum: -11506.0000\nIteration No: 14 started. Evaluating function at random point.\n[19, 47, 15, 357, 'log2']\n10128.0\nIteration No: 14 ended. Evaluation done at random point.\nTime taken: 8.2594\nFunction value obtained: -10128.0000\nCurrent minimum: -11506.0000\nIteration No: 15 started. Evaluating function at random point.\n[19, 46, 11, 521, 'log2']\n10282.0\nIteration No: 15 ended. Evaluation done at random point.\nTime taken: 7.9045\nFunction value obtained: -10282.0000\nCurrent minimum: -11506.0000\nIteration No: 16 started. Evaluating function at random point.\n[19, 47, 13, 573, 40]\n11452.0\nIteration No: 16 ended. Evaluation done at random point.\nTime taken: 19.3179\nFunction value obtained: -11452.0000\nCurrent minimum: -11506.0000\nIteration No: 17 started. Evaluating function at random point.\n[18, 47, 15, 518, 'log2']\n9932.0\nIteration No: 17 ended. Evaluation done at random point.\nTime taken: 7.8443\nFunction value obtained: -9932.0000\nCurrent minimum: -11506.0000\nIteration No: 18 started. Evaluating function at random point.\n[19, 47, 14, 511, 40]\n11234.0\nIteration No: 18 ended. Evaluation done at random point.\nTime taken: 20.0079\nFunction value obtained: -11234.0000\nCurrent minimum: -11506.0000\nIteration No: 19 started. Evaluating function at random point.\n[18, 46, 11, 520, 'auto']\n10860.0\nIteration No: 19 ended. Evaluation done at random point.\nTime taken: 11.3418\nFunction value obtained: -10860.0000\nCurrent minimum: -11506.0000\nIteration No: 20 started. Evaluating function at random point.\n[19, 46, 10, 460, 40]\n11066.0\nIteration No: 20 ended. Evaluation done at random point.\nTime taken: 20.5185\nFunction value obtained: -11066.0000\nCurrent minimum: -11506.0000\nIteration No: 21 started. Searching for the next optimal point.\n[18, 47, 12, 366, 40]\n11468.0\nIteration No: 21 ended. Search finished for the next optimal point.\nTime taken: 20.5306\nFunction value obtained: -11468.0000\nCurrent minimum: -11506.0000\nIteration No: 22 started. Searching for the next optimal point.\n[18, 46, 11, 544, 40]\n11268.0\nIteration No: 22 ended. Search finished for the next optimal point.\nTime taken: 19.3757\nFunction value obtained: -11268.0000\nCurrent minimum: -11506.0000\nIteration No: 23 started. Searching for the next optimal point.\n[19, 47, 14, 343, 40]\n11586.0\nIteration No: 23 ended. Search finished for the next optimal point.\nTime taken: 20.3535\nFunction value obtained: -11586.0000\nCurrent minimum: -11586.0000\nIteration No: 24 started. Searching for the next optimal point.\n[19, 46, 15, 316, 40]\n11296.0\nIteration No: 24 ended. Search finished for the next optimal point.\nTime taken: 19.5642\nFunction value obtained: -11296.0000\nCurrent minimum: -11586.0000\nIteration No: 25 started. Searching for the next optimal point.\n[19, 47, 11, 311, 40]\n11514.0\nIteration No: 25 ended. Search finished for the next optimal point.\nTime taken: 20.4922\nFunction value obtained: -11514.0000\nCurrent minimum: -11586.0000\nIteration No: 26 started. Searching for the next optimal point.\n[19, 47, 14, 344, 40]\n11632.0\nIteration No: 26 ended. Search finished for the next optimal point.\nTime taken: 19.8204\nFunction value obtained: -11632.0000\nCurrent minimum: -11632.0000\nIteration No: 27 started. Searching for the next optimal point.\n[19, 47, 10, 312, 40]\n11168.0\nIteration No: 27 ended. Search finished for the next optimal point.\nTime taken: 21.3242\nFunction value obtained: -11168.0000\nCurrent minimum: -11632.0000\nIteration No: 28 started. Searching for the next optimal point.\n[19, 47, 15, 306, 40]\n11406.0\nIteration No: 28 ended. Search finished for the next optimal point.\nTime taken: 20.7973\nFunction value obtained: -11406.0000\nCurrent minimum: -11632.0000\nIteration No: 29 started. Searching for the next optimal point.\n[19, 47, 15, 345, 40]\n11636.0\nIteration No: 29 ended. Search finished for the next optimal point.\nTime taken: 20.1946\nFunction value obtained: -11636.0000\nCurrent minimum: -11636.0000\nIteration No: 30 started. Searching for the next optimal point.\n[19, 47, 14, 344, 40]\n11632.0\nIteration No: 30 ended. Search finished for the next optimal point.\nTime taken: 20.3341\nFunction value obtained: -11632.0000\nCurrent minimum: -11636.0000\nIteration No: 31 started. Searching for the next optimal point.\n[19, 47, 14, 347, 40]\n11468.0\nIteration No: 31 ended. Search finished for the next optimal point.\nTime taken: 20.1338\nFunction value obtained: -11468.0000\nCurrent minimum: -11636.0000\nIteration No: 32 started. Searching for the next optimal point.\n[19, 47, 14, 345, 40]\n11472.0\nIteration No: 32 ended. Search finished for the next optimal point.\nTime taken: 20.2519\nFunction value obtained: -11472.0000\nCurrent minimum: -11636.0000\nIteration No: 33 started. Searching for the next optimal point.\n[19, 47, 14, 342, 40]\n11520.0\nIteration No: 33 ended. Search finished for the next optimal point.\nTime taken: 20.2577\nFunction value obtained: -11520.0000\nCurrent minimum: -11636.0000\nIteration No: 34 started. Searching for the next optimal point.\n[19, 47, 15, 340, 40]\n11418.0\nIteration No: 34 ended. Search finished for the next optimal point.\nTime taken: 20.0289\nFunction value obtained: -11418.0000\nCurrent minimum: -11636.0000\nIteration No: 35 started. Searching for the next optimal point.\n[19, 47, 14, 342, 40]\n11520.0\nIteration No: 35 ended. Search finished for the next optimal point.\nTime taken: 20.2006\nFunction value obtained: -11520.0000\nCurrent minimum: -11636.0000\nIteration No: 36 started. Searching for the next optimal point.\n[19, 47, 14, 346, 40]\n11534.0\nIteration No: 36 ended. Search finished for the next optimal point.\nTime taken: 20.3947\nFunction value obtained: -11534.0000\nCurrent minimum: -11636.0000\nIteration No: 37 started. Searching for the next optimal point.\n[19, 47, 10, 343, 40]\n11374.0\nIteration No: 37 ended. Search finished for the next optimal point.\nTime taken: 20.3695\nFunction value obtained: -11374.0000\nCurrent minimum: -11636.0000\nIteration No: 38 started. Searching for the next optimal point.\n[19, 47, 14, 315, 40]\n11620.0\nIteration No: 38 ended. Search finished for the next optimal point.\nTime taken: 20.4518\nFunction value obtained: -11620.0000\nCurrent minimum: -11636.0000\nIteration No: 39 started. Searching for the next optimal point.\n[19, 47, 14, 306, 40]\n11556.0\nIteration No: 39 ended. Search finished for the next optimal point.\nTime taken: 20.1591\nFunction value obtained: -11556.0000\nCurrent minimum: -11636.0000\nIteration No: 40 started. Searching for the next optimal point.\n[19, 47, 14, 320, 40]\n11408.0\nIteration No: 40 ended. Search finished for the next optimal point.\nTime taken: 19.7747\nFunction value obtained: -11408.0000\nCurrent minimum: -11636.0000\nIteration No: 41 started. Searching for the next optimal point.\n[19, 47, 14, 344, 40]\n11632.0\nIteration No: 41 ended. Search finished for the next optimal point.\nTime taken: 20.1468\nFunction value obtained: -11632.0000\nCurrent minimum: -11636.0000\nIteration No: 42 started. Searching for the next optimal point.\n[19, 47, 14, 353, 40]\n11448.0\nIteration No: 42 ended. Search finished for the next optimal point.\nTime taken: 19.5073\nFunction value obtained: -11448.0000\nCurrent minimum: -11636.0000\nIteration No: 43 started. Searching for the next optimal point.\n[19, 47, 14, 323, 40]\n11368.0\nIteration No: 43 ended. Search finished for the next optimal point.\nTime taken: 20.0986\nFunction value obtained: -11368.0000\nCurrent minimum: -11636.0000\nIteration No: 44 started. Searching for the next optimal point.\n[19, 47, 13, 344, 40]\n11592.0\nIteration No: 44 ended. Search finished for the next optimal point.\nTime taken: 20.0278\nFunction value obtained: -11592.0000\nCurrent minimum: -11636.0000\nIteration No: 45 started. Searching for the next optimal point.\n[19, 47, 14, 347, 40]\n11468.0\nIteration No: 45 ended. Search finished for the next optimal point.\nTime taken: 19.7845\nFunction value obtained: -11468.0000\nCurrent minimum: -11636.0000\nIteration No: 46 started. Searching for the next optimal point.\n[19, 47, 14, 344, 40]\n11632.0\nIteration No: 46 ended. Search finished for the next optimal point.\nTime taken: 19.9381\nFunction value obtained: -11632.0000\nCurrent minimum: -11636.0000\nIteration No: 47 started. Searching for the next optimal point.\n[19, 47, 14, 343, 40]\n11586.0\nIteration No: 47 ended. Search finished for the next optimal point.\nTime taken: 19.4966\nFunction value obtained: -11586.0000\nCurrent minimum: -11636.0000\nIteration No: 48 started. Searching for the next optimal point.\n[19, 47, 14, 345, 40]\n11472.0\nIteration No: 48 ended. Search finished for the next optimal point.\nTime taken: 20.0669\nFunction value obtained: -11472.0000\nCurrent minimum: -11636.0000\nIteration No: 49 started. Searching for the next optimal point.\n[19, 47, 14, 344, 40]\n11632.0\nIteration No: 49 ended. Search finished for the next optimal point.\nTime taken: 20.5108\nFunction value obtained: -11632.0000\nCurrent minimum: -11636.0000\nIteration No: 50 started. Searching for the next optimal point.\n[19, 47, 14, 487, 40]\n11236.0\nIteration No: 50 ended. Search finished for the next optimal point.\nTime taken: 19.3586\nFunction value obtained: -11236.0000\nCurrent minimum: -11636.0000\nCPU times: user 2min 13s, sys: 29.8 s, total: 2min 43s\nWall time: 14min 10s\n","output_type":"stream"}]},{"cell_type":"code","source":"param = ['max_depth        ', \n         'n_estimators     ', \n         'class_weight     ', \n         'min_samples_split', \n         'min_samples_split', \n         'max_features     '\n        ]\n\nfor i in range(0, len(result.x)): \n    print('{} \\t {}'.format(param[i], result.x[i]))","metadata":{"execution":{"iopub.status.busy":"2021-09-03T12:05:59.463283Z","iopub.execute_input":"2021-09-03T12:05:59.463684Z","iopub.status.idle":"2021-09-03T12:05:59.471269Z","shell.execute_reply.started":"2021-09-03T12:05:59.463652Z","shell.execute_reply":"2021-09-03T12:05:59.470012Z"},"trusted":true},"execution_count":22,"outputs":[{"name":"stdout","text":"max_depth         \t 19\nn_estimators      \t 47\nclass_weight      \t 15\nmin_samples_split \t 345\nmin_samples_split \t 40\n","output_type":"stream"}]},{"cell_type":"code","source":"model = RandomForestClassifier(\n                        max_depth = result.x[0],\n                        n_estimators = result.x[1],\n                        class_weight = {0:1,1:result.x[2]},\n                        min_samples_split = result.x[3],\n                        max_features = result.x[4],\n                        n_jobs=-1,\n                        random_state=1234\n        )\n\n\n\nmodel.fit(X_train[selected_features], y_train)\n\nbst_max_profit_model = model\n\nprint('treino:')\nshow_performance(bst_max_profit_model, X_train[selected_features], y_train)\n\nprint('test:')\nshow_performance(bst_max_profit_model, X_test[selected_features], y_test)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T12:06:04.449760Z","iopub.execute_input":"2021-09-03T12:06:04.450310Z","iopub.status.idle":"2021-09-03T12:06:09.922261Z","shell.execute_reply.started":"2021-09-03T12:06:04.450246Z","shell.execute_reply":"2021-09-03T12:06:09.921003Z"},"trusted":true},"execution_count":23,"outputs":[{"name":"stdout","text":"treino:\nMatriz de confusão:\n[[51512  6898]\n [  777  1629]]\n\nLucro obtido com a ação (em Reais): R$ 77630 \nLucro maximo possivel: R$ 216540 \nPorcentagem do lucro maximo: 35.85%\n\ntest:\nMatriz de confusão:\n[[12810  1792]\n [  233   369]]\n\nLucro obtido com a ação (em Reais): R$ 15290 \nLucro maximo possivel: R$ 54180 \nPorcentagem do lucro maximo: 28.22%\n\n","output_type":"stream"}]},{"cell_type":"code","source":"%%time\nskf = StratifiedKFold(n_splits=5, shuffle = True, random_state = 1234)\n\ngs_rf = GridSearchCV(estimator=RandomForestClassifier(random_state=1234, n_jobs=-1), \n                       param_grid = {'max_depth':[18],\n                                     'n_estimators':[47],\n                                     'criterion': ['gini'],\n                                     'class_weight':[{0:1,1:12}, {0:1,1:13}, {0:1,1:14}], \n                                     'min_samples_split':[int(X_train.shape[0]*0.01),\n                                                          int(X_train.shape[0]*0.005)],\n                                     'max_features':['auto','log2', 40]\n                                     }, \n                       scoring=max_profit_scorer, \n                       cv=skf.split(X_train[selected_features], y_train), \n                       n_jobs=-1, \n                       verbose=5)\n\ngs_rf.fit(X_train[selected_features], y_train)\n\n\nbest_rf = gs_rf.best_estimator_\n\nprint('treino:')\nshow_performance(best_rf, X_train[selected_features], y_train)\n\nprint('test:')\nshow_performance(best_rf, X_test[selected_features], y_test)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T01:44:20.159830Z","iopub.execute_input":"2021-09-03T01:44:20.160370Z","iopub.status.idle":"2021-09-03T01:48:20.166408Z","shell.execute_reply.started":"2021-09-03T01:44:20.160327Z","shell.execute_reply":"2021-09-03T01:48:20.165195Z"},"trusted":true},"execution_count":46,"outputs":[{"name":"stdout","text":"Fitting 5 folds for each of 18 candidates, totalling 90 fits\n","output_type":"stream"},{"name":"stderr","text":"[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.\n[Parallel(n_jobs=-1)]: Done  10 tasks      | elapsed:   27.6s\n[Parallel(n_jobs=-1)]: Done  64 tasks      | elapsed:  2.8min\n[Parallel(n_jobs=-1)]: Done  90 out of  90 | elapsed:  3.9min finished\n","output_type":"stream"},{"name":"stdout","text":"treino:\nMatriz de confusão:\n[[52009  6401]\n [  894  1512]]\n\nLucro obtido com a ação (em Reais): R$ 72070 \nLucro maximo possivel: R$ 216540 \nPorcentagem do lucro maximo: 33.28%\n\ntest:\nMatriz de confusão:\n[[12948  1654]\n [  246   356]]\n\nLucro obtido com a ação (em Reais): R$ 15500 \nLucro maximo possivel: R$ 54180 \nPorcentagem do lucro maximo: 28.61%\n\nCPU times: user 20.3 s, sys: 680 ms, total: 20.9 s\nWall time: 3min 59s\n","output_type":"stream"}]},{"cell_type":"markdown","source":"Só com a retirada de colunas sem variancia e duplicadas obtemos um lucro de 28.24% do lucro máximo possivel na partição de teste.","metadata":{}},{"cell_type":"markdown","source":"Agora procedemos a retirar as features com variancias próximas a zero.","metadata":{}},{"cell_type":"code","source":"qtd = get_uniques(X_train[selected_features])\nnear_zero_variance = qtd[qtd['concentration'] >= 0.995]\n#near_zero_variance.sort_values('concentration')","metadata":{"execution":{"iopub.status.busy":"2021-09-03T12:18:33.482234Z","iopub.execute_input":"2021-09-03T12:18:33.482636Z","iopub.status.idle":"2021-09-03T12:18:34.031670Z","shell.execute_reply.started":"2021-09-03T12:18:33.482603Z","shell.execute_reply":"2021-09-03T12:18:34.030431Z"},"trusted":true},"execution_count":24,"outputs":[]},{"cell_type":"code","source":"features_nzv = list(near_zero_variance['var'])\nlen(features_nzv)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T12:18:36.478574Z","iopub.execute_input":"2021-09-03T12:18:36.478918Z","iopub.status.idle":"2021-09-03T12:18:36.484480Z","shell.execute_reply.started":"2021-09-03T12:18:36.478887Z","shell.execute_reply":"2021-09-03T12:18:36.483824Z"},"trusted":true},"execution_count":25,"outputs":[{"execution_count":25,"output_type":"execute_result","data":{"text/plain":"149"},"metadata":{}}]},{"cell_type":"code","source":"selected_features = list(set(X_train.columns) \n                         - set(zero_variance) \n                         - set(duplicate_vars) \n                         - set(features_nzv)\n                        )\nlen(selected_features)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T12:18:44.403320Z","iopub.execute_input":"2021-09-03T12:18:44.403665Z","iopub.status.idle":"2021-09-03T12:18:44.410532Z","shell.execute_reply.started":"2021-09-03T12:18:44.403636Z","shell.execute_reply":"2021-09-03T12:18:44.409416Z"},"trusted":true},"execution_count":26,"outputs":[{"execution_count":26,"output_type":"execute_result","data":{"text/plain":"155"},"metadata":{}}]},{"cell_type":"code","source":"model.fit(X_train[selected_features], y_train)\n\nprint('treino:')\nshow_performance(model, X_train[selected_features], y_train)\n\nprint('test:')\nshow_performance(model, X_test[selected_features], y_test)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T12:18:52.701484Z","iopub.execute_input":"2021-09-03T12:18:52.701984Z","iopub.status.idle":"2021-09-03T12:18:59.201579Z","shell.execute_reply.started":"2021-09-03T12:18:52.701951Z","shell.execute_reply":"2021-09-03T12:18:59.200539Z"},"trusted":true},"execution_count":27,"outputs":[{"name":"stdout","text":"treino:\nMatriz de confusão:\n[[51211  7199]\n [  725  1681]]\n\nLucro obtido com a ação (em Reais): R$ 79300 \nLucro maximo possivel: R$ 216540 \nPorcentagem do lucro maximo: 36.62%\n\ntest:\nMatriz de confusão:\n[[12752  1850]\n [  229   373]]\n\nLucro obtido com a ação (em Reais): R$ 15070 \nLucro maximo possivel: R$ 54180 \nPorcentagem do lucro maximo: 27.81%\n\n","output_type":"stream"}]},{"cell_type":"markdown","source":"Apos a remoção de 149 features com variancia quase zero e mantendo os mesmos hiperparametros do modelo é observado uma diminuição da performance, resaltamos que dita diminuição não é alarme para voltar atrás, e alguns pontos devem ser discutidos. Primeiramente resaltamos que os hiperparametros se mantiveram constantes, tal vez um fine-tunning seria uma boa abordagem, isso porque é aconselhado que alguns hiperparametros sejam definidos em função do número de features (controle de overfittinng underfitting). Agora, uma outra sugestão seria criar uma nova feature pela combinação linear das variaveis removidas, isso a fim de manter a combinação das poucas variancias observadas nas features retiradas, isso também seria interessante principalmente para este tipo de problemas onde existe um grande desvalanceamento (vai que a pouca variancia retirada se encontre concentrada nas observações de classe possitiva).","metadata":{}},{"cell_type":"markdown","source":"Também retiramos as colunas \"quase-duplicadas\", isto é as que possuim um valor de correlação muito elevada. Neste casso utilizamos o calculo de correlação de Pearson e um thread de 0.95 (sendo que 1 é correlação perfeita = duplicada)","metadata":{}},{"cell_type":"code","source":"def get_redundant_pairs(df):\n    '''Get diagonal and lower triangular pairs of correlation matrix'''\n    pairs_to_drop = set()\n    cols = df.columns\n    for i in range(0, df.shape[1]):\n        for j in range(0, i+1):\n            pairs_to_drop.add((cols[i], cols[j]))\n    return pairs_to_drop\n\ndef get_top_abs_correlations(df, thread=0.95):\n    au_corr = df.corr().abs().unstack()\n    labels_to_drop = get_redundant_pairs(df)\n    au_corr = au_corr.drop(labels=labels_to_drop).sort_values(ascending=False)\n    df_result = pd.DataFrame(au_corr).reset_index()\n    df_result.columns = ['level_0', 'level_1', 'corr']\n    return df_result[df_result['corr'] > thread]\n\n#print(\"Top Absolute Correlations\")\nget_top_abs_correlations(X_train[selected_features], 0.95)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T12:19:20.688452Z","iopub.execute_input":"2021-09-03T12:19:20.688912Z","iopub.status.idle":"2021-09-03T12:19:25.118273Z","shell.execute_reply.started":"2021-09-03T12:19:20.688873Z","shell.execute_reply":"2021-09-03T12:19:25.117551Z"},"trusted":true},"execution_count":28,"outputs":[{"execution_count":28,"output_type":"execute_result","data":{"text/plain":"                         level_0                       level_1      corr\n0                       num_var8                      ind_var8  1.000000\n1      delta_imp_aport_var13_1y3     delta_num_aport_var13_1y3  1.000000\n2                     ind_var8_0                    num_var8_0  0.999742\n3                num_var13_corto               ind_var13_corto  0.999589\n4                      num_var24                     ind_var24  0.999554\n5        num_op_var39_efect_ult3       num_op_var41_efect_ult3  0.999294\n6                    ind_var40_0                   num_var40_0  0.999279\n7        imp_op_var39_efect_ult3       imp_op_var41_efect_ult3  0.999129\n8        num_op_var41_efect_ult1       num_op_var39_efect_ult1  0.998815\n9                     ind_var1_0                    num_var1_0  0.998570\n10                   ind_var40_0                    ind_var1_0  0.998554\n11       imp_op_var39_efect_ult1       imp_op_var41_efect_ult1  0.998025\n12                   num_var40_0                    num_var1_0  0.997849\n13                    ind_var1_0                   num_var40_0  0.997834\n14                num_var45_ult3            num_med_var45_ult3  0.997792\n15                   ind_var24_0                   num_var24_0  0.997626\n16                   ind_var40_0                    num_var1_0  0.997134\n17                   saldo_var12                   saldo_var24  0.996463\n18                     num_var14                     ind_var14  0.995556\n19                     num_var12                     ind_var12  0.994935\n20             ind_var13_corto_0             num_var13_corto_0  0.994637\n21                      ind_var5                      num_var5  0.994425\n22            num_op_var39_hace2            num_op_var41_hace2  0.994274\n23               ind_var13_largo             ind_var13_largo_0  0.993322\n24             num_op_var41_ult3             num_op_var39_ult3  0.990496\n25             imp_op_var41_ult1             imp_op_var39_ult1  0.990272\n26                   ind_var41_0                   ind_var39_0  0.990230\n27             saldo_var13_corto  saldo_medio_var13_corto_ult1  0.988434\n28  saldo_medio_var13_corto_ult1  saldo_medio_var13_corto_ult3  0.988427\n29             num_op_var41_ult1             num_op_var39_ult1  0.987834\n30                   saldo_var26                   saldo_var25  0.987013\n31            num_op_var39_hace3            num_op_var41_hace3  0.986951\n32  saldo_medio_var13_largo_ult1  saldo_medio_var13_largo_ult3  0.986662\n33        saldo_medio_var12_ult3        saldo_medio_var12_ult1  0.986631\n34                     ind_var13                   ind_var13_0  0.985164\n35                   saldo_var12        saldo_medio_var12_ult1  0.985050\n36                     num_var35                      num_var4  0.983122\n37                   saldo_var24        saldo_medio_var12_ult1  0.983112\n38             ind_var13_corto_0               ind_var13_corto  0.981517\n39               num_var13_corto             ind_var13_corto_0  0.981113\n40             saldo_var13_corto  saldo_medio_var13_corto_ult3  0.980582\n41                   num_var26_0                   num_var25_0  0.979067\n42                   ind_var25_0                   ind_var26_0  0.978909\n43                 ind_var25_cte                 ind_var26_cte  0.978395\n44               num_var13_largo             num_var13_largo_0  0.977765\n45                   num_var13_0                     num_var13  0.977522\n46               num_var13_corto             num_var13_corto_0  0.977032\n47             num_var13_corto_0               ind_var13_corto  0.976646\n48                   saldo_var12                   saldo_var42  0.975986\n49       num_op_var39_comer_ult1       num_op_var41_comer_ult1  0.975109\n50                   saldo_var12        saldo_medio_var12_ult3  0.972544\n51             ind_var10cte_ult1             ind_var9_cte_ult1  0.972390\n52                   saldo_var24                   saldo_var42  0.972368\n53                   saldo_var24        saldo_medio_var12_ult3  0.970993\n54                     ind_var13                     num_var13  0.969087\n55                 ind_var9_ult1                ind_var10_ult1  0.966877\n56                   saldo_var42        saldo_medio_var12_ult1  0.961314\n57       imp_op_var39_comer_ult3       imp_op_var41_comer_ult3  0.960993\n58             ind_var13_corto_0    num_meses_var13_corto_ult3  0.960424\n59       num_op_var39_comer_ult3       num_op_var41_comer_ult3  0.959424\n60       imp_op_var39_comer_ult1       imp_op_var41_comer_ult1  0.958701\n61                   num_var13_0                   ind_var13_0  0.958033\n62         saldo_medio_var8_ult1         saldo_medio_var8_ult3  0.957583\n63    num_meses_var13_corto_ult3               ind_var13_corto  0.956614\n64               num_var13_corto    num_meses_var13_corto_ult3  0.956110\n65             num_var13_corto_0    num_meses_var13_corto_ult3  0.956053\n66                   ind_var13_0                     num_var13  0.954710\n67                     ind_var12          num_meses_var12_ult3  0.951673\n68                   num_var41_0                   num_var39_0  0.951621","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>level_0</th>\n      <th>level_1</th>\n      <th>corr</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>num_var8</td>\n      <td>ind_var8</td>\n      <td>1.000000</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>delta_imp_aport_var13_1y3</td>\n      <td>delta_num_aport_var13_1y3</td>\n      <td>1.000000</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>ind_var8_0</td>\n      <td>num_var8_0</td>\n      <td>0.999742</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>num_var13_corto</td>\n      <td>ind_var13_corto</td>\n      <td>0.999589</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>num_var24</td>\n      <td>ind_var24</td>\n      <td>0.999554</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>num_op_var39_efect_ult3</td>\n      <td>num_op_var41_efect_ult3</td>\n      <td>0.999294</td>\n    </tr>\n    <tr>\n      <th>6</th>\n      <td>ind_var40_0</td>\n      <td>num_var40_0</td>\n      <td>0.999279</td>\n    </tr>\n    <tr>\n      <th>7</th>\n      <td>imp_op_var39_efect_ult3</td>\n      <td>imp_op_var41_efect_ult3</td>\n      <td>0.999129</td>\n    </tr>\n    <tr>\n      <th>8</th>\n      <td>num_op_var41_efect_ult1</td>\n      <td>num_op_var39_efect_ult1</td>\n      <td>0.998815</td>\n    </tr>\n    <tr>\n      <th>9</th>\n      <td>ind_var1_0</td>\n      <td>num_var1_0</td>\n      <td>0.998570</td>\n    </tr>\n    <tr>\n      <th>10</th>\n      <td>ind_var40_0</td>\n      <td>ind_var1_0</td>\n      <td>0.998554</td>\n    </tr>\n    <tr>\n      <th>11</th>\n      <td>imp_op_var39_efect_ult1</td>\n      <td>imp_op_var41_efect_ult1</td>\n      <td>0.998025</td>\n    </tr>\n    <tr>\n      <th>12</th>\n      <td>num_var40_0</td>\n      <td>num_var1_0</td>\n      <td>0.997849</td>\n    </tr>\n    <tr>\n      <th>13</th>\n      <td>ind_var1_0</td>\n      <td>num_var40_0</td>\n      <td>0.997834</td>\n    </tr>\n    <tr>\n      <th>14</th>\n      <td>num_var45_ult3</td>\n      <td>num_med_var45_ult3</td>\n      <td>0.997792</td>\n    </tr>\n    <tr>\n      <th>15</th>\n      <td>ind_var24_0</td>\n      <td>num_var24_0</td>\n      <td>0.997626</td>\n    </tr>\n    <tr>\n      <th>16</th>\n      <td>ind_var40_0</td>\n      <td>num_var1_0</td>\n      <td>0.997134</td>\n    </tr>\n    <tr>\n      <th>17</th>\n      <td>saldo_var12</td>\n      <td>saldo_var24</td>\n      <td>0.996463</td>\n    </tr>\n    <tr>\n      <th>18</th>\n      <td>num_var14</td>\n      <td>ind_var14</td>\n      <td>0.995556</td>\n    </tr>\n    <tr>\n      <th>19</th>\n      <td>num_var12</td>\n      <td>ind_var12</td>\n      <td>0.994935</td>\n    </tr>\n    <tr>\n      <th>20</th>\n      <td>ind_var13_corto_0</td>\n      <td>num_var13_corto_0</td>\n      <td>0.994637</td>\n    </tr>\n    <tr>\n      <th>21</th>\n      <td>ind_var5</td>\n      <td>num_var5</td>\n      <td>0.994425</td>\n    </tr>\n    <tr>\n      <th>22</th>\n      <td>num_op_var39_hace2</td>\n      <td>num_op_var41_hace2</td>\n      <td>0.994274</td>\n    </tr>\n    <tr>\n      <th>23</th>\n      <td>ind_var13_largo</td>\n      <td>ind_var13_largo_0</td>\n      <td>0.993322</td>\n    </tr>\n    <tr>\n      <th>24</th>\n      <td>num_op_var41_ult3</td>\n      <td>num_op_var39_ult3</td>\n      <td>0.990496</td>\n    </tr>\n    <tr>\n      <th>25</th>\n      <td>imp_op_var41_ult1</td>\n      <td>imp_op_var39_ult1</td>\n      <td>0.990272</td>\n    </tr>\n    <tr>\n      <th>26</th>\n      <td>ind_var41_0</td>\n      <td>ind_var39_0</td>\n      <td>0.990230</td>\n    </tr>\n    <tr>\n      <th>27</th>\n      <td>saldo_var13_corto</td>\n      <td>saldo_medio_var13_corto_ult1</td>\n      <td>0.988434</td>\n    </tr>\n    <tr>\n      <th>28</th>\n      <td>saldo_medio_var13_corto_ult1</td>\n      <td>saldo_medio_var13_corto_ult3</td>\n      <td>0.988427</td>\n    </tr>\n    <tr>\n      <th>29</th>\n      <td>num_op_var41_ult1</td>\n      <td>num_op_var39_ult1</td>\n      <td>0.987834</td>\n    </tr>\n    <tr>\n      <th>30</th>\n      <td>saldo_var26</td>\n      <td>saldo_var25</td>\n      <td>0.987013</td>\n    </tr>\n    <tr>\n      <th>31</th>\n      <td>num_op_var39_hace3</td>\n      <td>num_op_var41_hace3</td>\n      <td>0.986951</td>\n    </tr>\n    <tr>\n      <th>32</th>\n      <td>saldo_medio_var13_largo_ult1</td>\n      <td>saldo_medio_var13_largo_ult3</td>\n      <td>0.986662</td>\n    </tr>\n    <tr>\n      <th>33</th>\n      <td>saldo_medio_var12_ult3</td>\n      <td>saldo_medio_var12_ult1</td>\n      <td>0.986631</td>\n    </tr>\n    <tr>\n      <th>34</th>\n      <td>ind_var13</td>\n      <td>ind_var13_0</td>\n      <td>0.985164</td>\n    </tr>\n    <tr>\n      <th>35</th>\n      <td>saldo_var12</td>\n      <td>saldo_medio_var12_ult1</td>\n      <td>0.985050</td>\n    </tr>\n    <tr>\n      <th>36</th>\n      <td>num_var35</td>\n      <td>num_var4</td>\n      <td>0.983122</td>\n    </tr>\n    <tr>\n      <th>37</th>\n      <td>saldo_var24</td>\n      <td>saldo_medio_var12_ult1</td>\n      <td>0.983112</td>\n    </tr>\n    <tr>\n      <th>38</th>\n      <td>ind_var13_corto_0</td>\n      <td>ind_var13_corto</td>\n      <td>0.981517</td>\n    </tr>\n    <tr>\n      <th>39</th>\n      <td>num_var13_corto</td>\n      <td>ind_var13_corto_0</td>\n      <td>0.981113</td>\n    </tr>\n    <tr>\n      <th>40</th>\n      <td>saldo_var13_corto</td>\n      <td>saldo_medio_var13_corto_ult3</td>\n      <td>0.980582</td>\n    </tr>\n    <tr>\n      <th>41</th>\n      <td>num_var26_0</td>\n      <td>num_var25_0</td>\n      <td>0.979067</td>\n    </tr>\n    <tr>\n      <th>42</th>\n      <td>ind_var25_0</td>\n      <td>ind_var26_0</td>\n      <td>0.978909</td>\n    </tr>\n    <tr>\n      <th>43</th>\n      <td>ind_var25_cte</td>\n      <td>ind_var26_cte</td>\n      <td>0.978395</td>\n    </tr>\n    <tr>\n      <th>44</th>\n      <td>num_var13_largo</td>\n      <td>num_var13_largo_0</td>\n      <td>0.977765</td>\n    </tr>\n    <tr>\n      <th>45</th>\n      <td>num_var13_0</td>\n      <td>num_var13</td>\n      <td>0.977522</td>\n    </tr>\n    <tr>\n      <th>46</th>\n      <td>num_var13_corto</td>\n      <td>num_var13_corto_0</td>\n      <td>0.977032</td>\n    </tr>\n    <tr>\n      <th>47</th>\n      <td>num_var13_corto_0</td>\n      <td>ind_var13_corto</td>\n      <td>0.976646</td>\n    </tr>\n    <tr>\n      <th>48</th>\n      <td>saldo_var12</td>\n      <td>saldo_var42</td>\n      <td>0.975986</td>\n    </tr>\n    <tr>\n      <th>49</th>\n      <td>num_op_var39_comer_ult1</td>\n      <td>num_op_var41_comer_ult1</td>\n      <td>0.975109</td>\n    </tr>\n    <tr>\n      <th>50</th>\n      <td>saldo_var12</td>\n      <td>saldo_medio_var12_ult3</td>\n      <td>0.972544</td>\n    </tr>\n    <tr>\n      <th>51</th>\n      <td>ind_var10cte_ult1</td>\n      <td>ind_var9_cte_ult1</td>\n      <td>0.972390</td>\n    </tr>\n    <tr>\n      <th>52</th>\n      <td>saldo_var24</td>\n      <td>saldo_var42</td>\n      <td>0.972368</td>\n    </tr>\n    <tr>\n      <th>53</th>\n      <td>saldo_var24</td>\n      <td>saldo_medio_var12_ult3</td>\n      <td>0.970993</td>\n    </tr>\n    <tr>\n      <th>54</th>\n      <td>ind_var13</td>\n      <td>num_var13</td>\n      <td>0.969087</td>\n    </tr>\n    <tr>\n      <th>55</th>\n      <td>ind_var9_ult1</td>\n      <td>ind_var10_ult1</td>\n      <td>0.966877</td>\n    </tr>\n    <tr>\n      <th>56</th>\n      <td>saldo_var42</td>\n      <td>saldo_medio_var12_ult1</td>\n      <td>0.961314</td>\n    </tr>\n    <tr>\n      <th>57</th>\n      <td>imp_op_var39_comer_ult3</td>\n      <td>imp_op_var41_comer_ult3</td>\n      <td>0.960993</td>\n    </tr>\n    <tr>\n      <th>58</th>\n      <td>ind_var13_corto_0</td>\n      <td>num_meses_var13_corto_ult3</td>\n      <td>0.960424</td>\n    </tr>\n    <tr>\n      <th>59</th>\n      <td>num_op_var39_comer_ult3</td>\n      <td>num_op_var41_comer_ult3</td>\n      <td>0.959424</td>\n    </tr>\n    <tr>\n      <th>60</th>\n      <td>imp_op_var39_comer_ult1</td>\n      <td>imp_op_var41_comer_ult1</td>\n      <td>0.958701</td>\n    </tr>\n    <tr>\n      <th>61</th>\n      <td>num_var13_0</td>\n      <td>ind_var13_0</td>\n      <td>0.958033</td>\n    </tr>\n    <tr>\n      <th>62</th>\n      <td>saldo_medio_var8_ult1</td>\n      <td>saldo_medio_var8_ult3</td>\n      <td>0.957583</td>\n    </tr>\n    <tr>\n      <th>63</th>\n      <td>num_meses_var13_corto_ult3</td>\n      <td>ind_var13_corto</td>\n      <td>0.956614</td>\n    </tr>\n    <tr>\n      <th>64</th>\n      <td>num_var13_corto</td>\n      <td>num_meses_var13_corto_ult3</td>\n      <td>0.956110</td>\n    </tr>\n    <tr>\n      <th>65</th>\n      <td>num_var13_corto_0</td>\n      <td>num_meses_var13_corto_ult3</td>\n      <td>0.956053</td>\n    </tr>\n    <tr>\n      <th>66</th>\n      <td>ind_var13_0</td>\n      <td>num_var13</td>\n      <td>0.954710</td>\n    </tr>\n    <tr>\n      <th>67</th>\n      <td>ind_var12</td>\n      <td>num_meses_var12_ult3</td>\n      <td>0.951673</td>\n    </tr>\n    <tr>\n      <th>68</th>\n      <td>num_var41_0</td>\n      <td>num_var39_0</td>\n      <td>0.951621</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"markdown","source":"Observamos 69 pares de variaveis altamente correlacionadas. Isso não significa que iremos a excluir 68, isso é porque podem existir correlações cruzada (3 ou mais variaveis altamente correlacionadas que geram combinações de pares, neste casso excluimos todas exceto 1).","metadata":{}},{"cell_type":"code","source":"corr_heatmap(X_train[['num_var13_corto', 'num_var13_corto_0', 'ind_var13_corto', 'num_meses_var13_corto_ult3']], \n             figsize_1= 10, \n             figsize_2 = 10, \n             see_numbers= True\n            )\n#Exemplo de correlação cruzada","metadata":{"execution":{"iopub.status.busy":"2021-09-03T12:19:40.400875Z","iopub.execute_input":"2021-09-03T12:19:40.401385Z","iopub.status.idle":"2021-09-03T12:19:40.814548Z","shell.execute_reply.started":"2021-09-03T12:19:40.401353Z","shell.execute_reply":"2021-09-03T12:19:40.813721Z"},"trusted":true},"execution_count":29,"outputs":[{"execution_count":29,"output_type":"execute_result","data":{"text/plain":"<AxesSubplot:>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<Figure size 720x720 with 2 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"A seguir calculamos a lista de features a ser excluida, sendo um total de 51.","metadata":{}},{"cell_type":"code","source":"cor_matrix = X_train[selected_features].corr().abs()\nupper_tri = cor_matrix.where(np.triu(np.ones(cor_matrix.shape),k=1).astype(np.bool))\nhight_corr = [column for column in upper_tri.columns if any(upper_tri[column] > 0.95)]\nprint(hight_corr)\nprint(len(hight_corr))","metadata":{"execution":{"iopub.status.busy":"2021-09-03T12:19:45.191666Z","iopub.execute_input":"2021-09-03T12:19:45.192031Z","iopub.status.idle":"2021-09-03T12:19:49.207940Z","shell.execute_reply.started":"2021-09-03T12:19:45.191999Z","shell.execute_reply":"2021-09-03T12:19:49.207030Z"},"trusted":true},"execution_count":30,"outputs":[{"name":"stdout","text":"['num_var5', 'ind_var13_0', 'ind_var1_0', 'ind_var12', 'imp_op_var41_comer_ult3', 'delta_num_aport_var13_1y3', 'ind_var13_corto_0', 'num_op_var39_ult3', 'imp_op_var41_efect_ult3', 'saldo_var24', 'saldo_medio_var12_ult3', 'num_var25_0', 'saldo_medio_var13_corto_ult1', 'num_op_var41_hace2', 'ind_var39_0', 'num_var8_0', 'ind_var9_cte_ult1', 'saldo_medio_var13_corto_ult3', 'num_var4', 'num_op_var39_ult1', 'num_var13_corto_0', 'imp_op_var39_ult1', 'ind_var13_largo_0', 'num_med_var45_ult3', 'ind_var14', 'num_meses_var12_ult3', 'ind_var8', 'num_var40_0', 'num_op_var39_efect_ult1', 'num_op_var41_comer_ult3', 'num_var39_0', 'num_var24_0', 'num_meses_var13_corto_ult3', 'saldo_medio_var13_largo_ult3', 'saldo_medio_var8_ult3', 'saldo_var25', 'ind_var24', 'num_op_var41_hace3', 'imp_op_var41_efect_ult1', 'num_op_var41_efect_ult3', 'saldo_var42', 'num_var13', 'num_var1_0', 'num_op_var41_comer_ult1', 'ind_var26_0', 'imp_op_var41_comer_ult1', 'ind_var13_corto', 'ind_var26_cte', 'saldo_medio_var12_ult1', 'ind_var10_ult1', 'num_var13_largo_0']\n51\n","output_type":"stream"}]},{"cell_type":"code","source":"selected_features = list(\n                        set(X_train.columns) \n                        - set(zero_variance) \n                        - set(duplicate_vars)\n                        - set(features_nzv) \n                        - set(hight_corr)\n                        )\nlen(selected_features)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:02:53.388273Z","iopub.execute_input":"2021-09-03T19:02:53.388827Z","iopub.status.idle":"2021-09-03T19:02:53.395945Z","shell.execute_reply.started":"2021-09-03T19:02:53.388787Z","shell.execute_reply":"2021-09-03T19:02:53.395028Z"},"trusted":true},"execution_count":142,"outputs":[{"execution_count":142,"output_type":"execute_result","data":{"text/plain":"104"},"metadata":{}}]},{"cell_type":"code","source":"model.fit(X_train[selected_features], y_train)\n\nprint('treino:')\nshow_performance(model, X_train[selected_features], y_train)\n\nprint('test:')\nshow_performance(model, X_test[selected_features], y_test)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T12:51:26.506117Z","iopub.execute_input":"2021-09-03T12:51:26.506587Z","iopub.status.idle":"2021-09-03T12:51:33.126752Z","shell.execute_reply.started":"2021-09-03T12:51:26.506546Z","shell.execute_reply":"2021-09-03T12:51:33.125045Z"},"trusted":true},"execution_count":35,"outputs":[{"name":"stdout","text":"treino:\nMatriz de confusão:\n[[51308  7102]\n [  725  1681]]\n\nLucro obtido com a ação (em Reais): R$ 80270 \nLucro maximo possivel: R$ 216540 \nPorcentagem do lucro maximo: 37.07%\n\ntest:\nMatriz de confusão:\n[[12770  1832]\n [  229   373]]\n\nLucro obtido com a ação (em Reais): R$ 15250 \nLucro maximo possivel: R$ 54180 \nPorcentagem do lucro maximo: 28.15%\n\n","output_type":"stream"}]},{"cell_type":"markdown","source":"Agora que sabemos que dita audacia (remoção de colunas correlacionadas) foi uma ação que melhorou o resultado do nosso modelo temos que dar uma pausa e ser sinceros, o calculo da correlação de Pearson, conforme calculado pelo .corr(), não garante a sua funcionalidade em variaves categoricas. Na execução anterior não fizemos distinção entre variaveis categoricas ou continuas, e mesmo assim conseguimos um melhor resultado (sorte nossa rsrs).\n\nComo descargo de consiencia, na seção Extra C apresentamos o resultado de treinar e testar o modelo apos alguns tratamento das variaveis categoricas como é a applicação do método get_dummies().\n\nPara o calculo da correlação em variaveis categoricas é sugerido o test de hipotese chi-quadrado.","metadata":{}},{"cell_type":"markdown","source":"Cabe resaltar que, mesmo mantendo os parametros do primeiro tunning e apos ter diminuido exageradamente o número de features, o modelo mantem uma performance aproximada ao primeiro teste, o qual sugere que as remoções guiadas por insights conceituais fazem sentido.","metadata":{}},{"cell_type":"markdown","source":"Agora testemos mais métodos de seleção de features, existe um metodo bastante conhecido e pratico que consite de adicionar uma feature completamente aleatorea ao nosso dataset para posteriormente treinar o modelo e aplicar a função feture_importance, seguido de rankear as variaveis de maior importancia e descartar todas aquelas cuja importancia seja menor à importancia da variavel randomica (ver trabalho de CaioMar). Embora seja um metodo poderoso e pouco custoso desde o ponto de vista computacional, neste casso não aplicaremos ele só para mostrar que também existem outros métodos de seleção de features rsrsrs (vai que Caio ache que estou copiando o trabalho dele).","metadata":{}},{"cell_type":"markdown","source":"O seguinte é um dos métodos de seleção de features univariantes, e funcionam pela seleção das melhores features baseadas em provas estatisticas univariadas como ANOVA. \n\nNeste caso utilisaremos o score f_classif, o qual é um método baseados no F-test o qual estima o grau de dependencia linear entre a feature e a variavel resposta... Sim asumiremos que as variaveis se relacionam linearmente o qual pode ser desvantagem.... Outra desvantagem é que este método também asume que as variaveis possuim uma distribución gaussiana o qual pode não ser verdade...","metadata":{}},{"cell_type":"markdown","source":"O seguinte teste consite de incrementar gradualmente o número de K features selecionadas para treinar nosso modelo, e a cada seleção de feature é realizado um tunning de hiperparametros pela aplicação de GridSearchCV (não existe motivo para ter mudado otimização bayesiana por random GridSearchCV, funcionaria igual de bem). Neste caso implicitamente a função GridSearchCV utiliza o método K-folds cross validation. Isso significa que para cada ponto do Grid são treinados K-1 modelos e testa sobre o fold restante. Como estamos com um dataset desbalanceado é importante utilizar a função StratifiedKFold o qual garante que a separação entre cada fold possua proporções iguais (ou muito aproximadas) de cada classe.","metadata":{}},{"cell_type":"markdown","source":"Aqui seria uma boa oportunidade para definir nossos parametros do grid em função do K casso se queira realizar um teste mais exaustivo da exploração dos hiperparametros a cada K features selecionadas.","metadata":{}},{"cell_type":"code","source":"%%time\n\nfrom sklearn.feature_selection import SelectKBest\nfrom sklearn.feature_selection import f_classif  \nimport copy\n\nX_train_clean = X_train[selected_features].copy()\nX_test_clean = X_test[selected_features].copy()\n\n\nK_vs_score_fc_train = [] #List to store AUC of each K with f_classif\nK_vs_score_fc_test = [] #List to store AUC of each K with f_classif\nK_vs_score_mic = [] #List to store AUC of each K with mutual_info_classif\n\n# save\nK_features = []\nK_selector = []\nK_estimator = []\n\nfor k in range(2, len(selected_features)+1, 2):\n    start = time()\n    \n    # Instanciamos nosso selector\n    selector_fc = SelectKBest(score_func = f_classif, k = k)\n    # Fitamos nosso selector\n    selector_fitted = selector_fc.fit(X_train_clean, y_train)\n    \n    # Extraimos as K features\n    cols = selector_fitted.get_support(indices=True)\n    features_df_new = X_train_clean.iloc[:,cols]\n\n    # Salvamos as K features selecionadas\n    K_features.append(list(features_df_new.columns))\n\n    # Transformamos nosso dataset\n    X_train_selected_fc = selector_fc.transform(X_train_clean)\n    \n    # salvamos nosso selector\n    K_selector.append(copy.deepcopy(selector_fc))\n\n\n    # Utilizamos estratificação na validação cruzada\n    skf = StratifiedKFold(n_splits=5, shuffle = True, random_state = 1234)\n    \n\n    # Aqui seria uma boa oportunidade para definir nossos parametros do grid em função do k\n    gsearch = GridSearchCV(estimator=RandomForestClassifier(random_state=1234, n_jobs=-1), \n                           param_grid = {'max_depth':[18],\n                                         'n_estimators':[47],\n                                         'criterion': ['gini'],\n                                         'class_weight':[{0:1,1:12}, {0:1,1:13}, {0:1,1:14}], \n                                         'min_samples_split':[int(X_train.shape[0]*0.01),\n                                                              int(X_train.shape[0]*0.005)],\n                                         'max_features':['auto','log2',40]\n                                         }, \n                           scoring=max_profit_scorer, \n                           cv=skf.split(X_train_selected_fc, y_train), \n                           n_jobs=-1, \n                           verbose=0)\n    \n    gsearch.fit(X_train_selected_fc, y_train)\n    bst_max_profit_model = gsearch.best_estimator_\n    K_estimator.append(bst_max_profit_model)\n    \n    y_train_pred = bst_max_profit_model.predict(X_train_selected_fc)\n    y_test_pred = bst_max_profit_model.predict(selector_fc.transform(X_test_clean))\n\n    y_train_optime = max_profit_score(y_train, y_train)\n    y_test_optime = max_profit_score(y_test, y_test)\n    \n    K_vs_score_fc_train.append(100*max_profit_score(y_train, y_train_pred)/y_train_optime)\n    K_vs_score_fc_test.append(100*max_profit_score(y_test, y_test_pred)/y_test_optime)\n    \n\n    end = time()\n    # Returning the metrics related to the tested K and the time spent on this iteration of the loop\n    print(\"index {}: -  k = {} - lucro_train = {} - lucro_test = {} - Time = {}s\".format(len(K_vs_score_fc_train)-1, k, K_vs_score_fc_train[-1], K_vs_score_fc_test[-1], end-start))","metadata":{"execution":{"iopub.status.busy":"2021-09-03T13:38:05.355803Z","iopub.execute_input":"2021-09-03T13:38:05.356230Z","iopub.status.idle":"2021-09-03T15:42:46.144573Z","shell.execute_reply.started":"2021-09-03T13:38:05.356192Z","shell.execute_reply":"2021-09-03T15:42:46.143357Z"},"trusted":true},"execution_count":39,"outputs":[{"name":"stdout","text":"index 0: -  k = 2 - lucro_train = -7.4628244204304055 - lucro_test = -7.438169066076043 - Time = 17.57846736907959s\nindex 1: -  k = 4 - lucro_train = -7.4304978295003234 - lucro_test = -7.881136950904393 - Time = 17.581249237060547s\nindex 2: -  k = 6 - lucro_train = 21.700378682922324 - lucro_test = 21.00406053894426 - Time = 26.152140617370605s\nindex 3: -  k = 8 - lucro_train = 23.7923709245405 - lucro_test = 22.019195275009228 - Time = 29.638583421707153s\nindex 4: -  k = 10 - lucro_train = 22.998060404544194 - lucro_test = 23.311184939091916 - Time = 31.974842071533203s\nindex 5: -  k = 12 - lucro_train = 27.72236076475478 - lucro_test = 25.987449243263196 - Time = 36.74688196182251s\nindex 6: -  k = 14 - lucro_train = 26.73409069917798 - lucro_test = 25.009228497600592 - Time = 36.814011096954346s\nindex 7: -  k = 16 - lucro_train = 27.833194790800775 - lucro_test = 25.378368401624215 - Time = 39.0928897857666s\nindex 8: -  k = 18 - lucro_train = 27.94402881684677 - lucro_test = 24.40014765596161 - Time = 40.4392306804657s\nindex 9: -  k = 20 - lucro_train = 27.847049044056526 - lucro_test = 24.953857511997047 - Time = 40.53828501701355s\nindex 10: -  k = 22 - lucro_train = 28.52128936916967 - lucro_test = 25.489110372831302 - Time = 41.39540481567383s\nindex 11: -  k = 24 - lucro_train = 27.58843631661587 - lucro_test = 26.33813215208564 - Time = 43.38228487968445s\nindex 12: -  k = 26 - lucro_train = 29.472614759397803 - lucro_test = 26.541159099298635 - Time = 45.48160767555237s\nindex 13: -  k = 28 - lucro_train = 29.463378590560634 - lucro_test = 26.873385012919897 - Time = 46.13991355895996s\nindex 14: -  k = 30 - lucro_train = 28.43816384963517 - lucro_test = 26.116648209671464 - Time = 46.76879954338074s\nindex 15: -  k = 32 - lucro_train = 27.948646901265356 - lucro_test = 26.356589147286822 - Time = 51.239405155181885s\nindex 16: -  k = 34 - lucro_train = 29.352544564514638 - lucro_test = 26.78110003691399 - Time = 50.123353719711304s\nindex 17: -  k = 36 - lucro_train = 29.42181583079339 - lucro_test = 27.150239940937617 - Time = 52.008918046951294s\nindex 18: -  k = 38 - lucro_train = 29.79126258428004 - lucro_test = 26.17201919527501 - Time = 50.58934926986694s\nindex 19: -  k = 40 - lucro_train = 34.59407037960654 - lucro_test = 26.393503137689184 - Time = 168.2811985015869s\nindex 20: -  k = 42 - lucro_train = 33.56423755426249 - lucro_test = 27.888519748984866 - Time = 170.62065029144287s\nindex 21: -  k = 44 - lucro_train = 31.204396416366492 - lucro_test = 27.334809892949426 - Time = 180.18634605407715s\nindex 22: -  k = 46 - lucro_train = 34.123025768911056 - lucro_test = 29.58656330749354 - Time = 180.62952995300293s\nindex 23: -  k = 48 - lucro_train = 33.84594070379607 - lucro_test = 29.032853451458102 - Time = 180.63824367523193s\nindex 24: -  k = 50 - lucro_train = 34.98198947076752 - lucro_test = 29.42045035068291 - Time = 182.00703477859497s\nindex 25: -  k = 52 - lucro_train = 31.763184631015054 - lucro_test = 28.71908453303802 - Time = 188.59941220283508s\nindex 26: -  k = 54 - lucro_train = 32.100304793571624 - lucro_test = 28.626799557032115 - Time = 191.13780856132507s\nindex 27: -  k = 56 - lucro_train = 33.55038330100674 - lucro_test = 28.054632705795495 - Time = 194.82574796676636s\nindex 28: -  k = 58 - lucro_train = 34.14611619100397 - lucro_test = 29.328165374677003 - Time = 195.84777116775513s\nindex 29: -  k = 60 - lucro_train = 37.33259443982636 - lucro_test = 28.516057585825028 - Time = 206.11400175094604s\nindex 30: -  k = 62 - lucro_train = 37.17557956959453 - lucro_test = 28.70062753783684 - Time = 206.5888352394104s\nindex 31: -  k = 64 - lucro_train = 37.68356885563868 - lucro_test = 27.260981912144704 - Time = 211.4976077079773s\nindex 32: -  k = 66 - lucro_train = 33.28253440472892 - lucro_test = 28.165374677002585 - Time = 207.0392370223999s\nindex 33: -  k = 68 - lucro_train = 34.15535235984114 - lucro_test = 28.71908453303802 - Time = 211.08215880393982s\nindex 34: -  k = 70 - lucro_train = 33.21326313845017 - lucro_test = 28.571428571428573 - Time = 209.63261437416077s\nindex 35: -  k = 72 - lucro_train = 37.25870508912903 - lucro_test = 28.294573643410853 - Time = 210.04832410812378s\nindex 36: -  k = 74 - lucro_train = 37.498845478895355 - lucro_test = 28.71908453303802 - Time = 206.11847829818726s\nindex 37: -  k = 76 - lucro_train = 37.6235337581971 - lucro_test = 28.737541528239202 - Time = 210.40435934066772s\nindex 38: -  k = 78 - lucro_train = 37.309504017733445 - lucro_test = 29.789590254706535 - Time = 209.68008589744568s\nindex 39: -  k = 80 - lucro_train = 38.16384963517133 - lucro_test = 29.863418235511258 - Time = 210.8973252773285s\nindex 40: -  k = 82 - lucro_train = 37.90061882331209 - lucro_test = 28.09154669619786 - Time = 217.54849457740784s\nindex 41: -  k = 84 - lucro_train = 38.2285028170315 - lucro_test = 28.922111480251015 - Time = 212.94809341430664s\nindex 42: -  k = 86 - lucro_train = 38.210030479357165 - lucro_test = 29.16205241786637 - Time = 213.93300080299377s\nindex 43: -  k = 88 - lucro_train = 38.210030479357165 - lucro_test = 28.70062753783684 - Time = 214.47555661201477s\nindex 44: -  k = 90 - lucro_train = 38.302392167728826 - lucro_test = 28.183831672203766 - Time = 219.6683051586151s\nindex 45: -  k = 92 - lucro_train = 38.34857301191466 - lucro_test = 28.571428571428573 - Time = 217.02458238601685s\nindex 46: -  k = 94 - lucro_train = 38.08072411563683 - lucro_test = 28.42377260981912 - Time = 212.68089652061462s\nindex 47: -  k = 96 - lucro_train = 37.85905606354484 - lucro_test = 28.922111480251015 - Time = 216.84857201576233s\nindex 48: -  k = 98 - lucro_train = 38.62565807702965 - lucro_test = 28.86674049464747 - Time = 214.59149503707886s\nindex 49: -  k = 100 - lucro_train = 38.74111018749423 - lucro_test = 28.202288667404947 - Time = 218.75727367401123s\nindex 50: -  k = 102 - lucro_train = 38.52406021982082 - lucro_test = 28.276116648209673 - Time = 221.7325496673584s\nindex 51: -  k = 104 - lucro_train = 38.6118038237739 - lucro_test = 27.427094868955333 - Time = 224.93333077430725s\nCPU times: user 10min 2s, sys: 8.34 s, total: 10min 11s\nWall time: 2h 4min 40s\n","output_type":"stream"}]},{"cell_type":"code","source":"fig, ax = plt.subplots(figsize = (20, 8))\nplt.plot(range(2, len(selected_features)+1, 2), K_vs_score_fc_train)\nplt.plot(range(2, len(selected_features)+1, 2), K_vs_score_fc_test)\nplt.title('Score para cada K features escolhidas', fontsize = 18)\nplt.xlabel('K', fontsize = 16)\nplt.xticks(range(2, len(selected_features)+1, 2))\nplt.ylabel('Score', fontsize = 16) \nax.spines['top'].set_visible(False)\nax.spines['right'].set_visible(False)\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-09-03T15:42:58.059251Z","iopub.execute_input":"2021-09-03T15:42:58.059699Z","iopub.status.idle":"2021-09-03T15:42:59.366431Z","shell.execute_reply.started":"2021-09-03T15:42:58.059657Z","shell.execute_reply":"2021-09-03T15:42:59.365256Z"},"trusted":true},"execution_count":40,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1440x576 with 1 Axes>","image/png":"iVBORw0KGgoAAAANSUhEUgAABJ4AAAH7CAYAAACNElWkAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8rg+JYAAAACXBIWXMAAAsTAAALEwEAmpwYAACajUlEQVR4nOzdd3xT573H8c+jZct7Y8yehxEggUBCErLInk2TKk2bNF1J0713c9umvb29Hbe7TXezG2U0e5OQvXcCCMIGA96W95Ce+8cRxoABG2RLhu/79fJL0tHRc36SZWN9eZ7fMdZaREREREREREREks2T6gJEREREREREROTgpOBJREREREREREQGhYInEREREREREREZFAqeRERERERERERkUCh4EhERERERERGRQaHgSUREREREREREBoWCJxERETnkOI7zL8dxbKrr2BPHcSY4jnOX4zjVjuNYx3H+leqaJLn6eg8O5H3pOM74xHvjB/t7PBERkaHgS3UBIiIiQ8lxnInAt4DjgbFAB7AVeAn4VyQSeSKF5ckw5zjOUuDISCSSs8v2AHAj8AHgD8DnI5HI3kKAfwGzgf/GfX+uHox6e9X3JaAhEon8azCPIyIiIoceBU8iInLIcBznSOBJoAu4HngXCAJTgNOAJkDBkySV4zhZwJ3A6cCPI5HI1fvYPwNYBPw+Eon8YghKBPgSsA438JLhYT3u76/uVBciIiKyNwqeRETkUPJ9IAs4PBKJvLnrnY7jlA99SeA4Tm4kEmlKxbEPhOM4fsAbiUTaU11LunIcpwC4H1gIfDkSify6Hw8bARigbvAqGzp6nwyOxIw5vaYiIpL2FDyJiMihZApQ21foBBCJRLbuus1xnJOArwFHA9lAJe6sqG9GIpGaxD4+4KvA5cBEoAV4CvivSCTydq+xxgNrgR8Cy4FvADOAW4GPJvY5JbF9AZAJrAT+GIlErt3Xk9tl/AjwbWAqUAX8A3e2TXev/acBXwBOwF126E3U9adIJPK3Xcb+AW5wdxjwCSAEjAQWA0sdx7kY+DBwOG5w0gQ8k3gN3tpX7b2OcyHw+cQ4AWAj8DDwtUgk0uk4jifxvE5PPLci3KVo9wPfi0QitbuMlwn8KFFbIfA28L09HHsB8BngGGA0EAPeAn4RiUT+09/n0Gu8EYnaDwM+3p9lbIleTpcnbn7fcZzvJ66fFIlElib2uRj3NZqD+z17G/h5JBK5fZex+vU96dX3Z9wuPYAmRCKRdYlt10UikY/uMv5HgX/uUtsP2Pv7JAP3Z+XDwCTc4OTpRE2v9xrbg/ve/DgwAbDAlkT9V0Uika69vY4DfJ3Oxv2Zm4n7M14NvAJ8KxKJrOy1XznwHeAcYBTQCLwJ/CwSiTzaa7/jgatxf4YDuD9Tf4hEIn/fV829xsgHfgpcCOQBrwJfiUQiL/baZzyJn/dIJPKDXtsH5T3vOM4Y3N8ti4HyxPN/D/hzJBK5rr/PTUREDj1qLi4iIoeS1UCx4zjv78/OjuN8CliC22vnT7gfYm8C5uF+SNvuJtwPiZuArwPXAicBzzuOc0QfQ78vMd5DuB+uH0wc70rgESAHt7fPVxI1/8lxnJ8P4Hmelxj/3kQ9Edww4K+77Hcibq+r+xL7XY27DPGvjuN8ew9j34Q7e+eXuAHClsT2zwFx4C/AZxPHWgQ86zjOlP4U7TjOfwO3A6XAr3CXf90FnIU7Uw3cD/JfB1YBP8d9/R7FDTmWJnop9XYLbnD4SuLyGdxlb/P6KOECYBoQBr6I+z0oAu50HOdD/XkOvZ7LONxAZRrwgQH0Tvoz8OXE9f8AlyW+lifG/THwb9wQ6WrcfmWtwG2O43x2l7H6+z25DKgBVvQ63mW4Acz+2u19kpj59BDue/H5xPP8KW74+mxiKex238V9D6wDvon7Pf9PYsyMfR28v6+T4zgnAPcABcD/4L5mfwWKgcm99huPG/58BliaqP3nQBQ4pdd+5wKPA9MTz/07uD9Tf0u8v/vrYdzfMdck6joMuN9xnNx+PDbp7/lEuP4obo+yf+O+Dj/FDcYXDeB5iYjIIUgznkRE5FDyY+BU4A7HcVbhfiB7GVgaiUSW997RcZzRwG9xP4wfE4lEGnrdfXViRgaO45yKO6sjDHxwe8Nox3HCuB9Uf8vuH8xmArN7H9NxnJGJff8diUR6hxx/dBznN8BXHMf5UyQSWdOP5zkHmB+JRF5LjP173A+eH3Uc58+RSOSFxH437DqTynGcX+F+cP6W4zi/6GNmSQNwSu+ZUwlnRCKRll3Guh54A/dD+mf2VnBi5sV3cGeTndV7WZbjON/qtWsHMDISibT12nat4zjPAX/DDfXCicedlri902wdx3Gewg0xdvXjSCSyU+DmOM5vgddxZ4zcvLfn0EsA971VAJwdiUSW9PNxRCKR5x3H2YIburwViURu7FXLXNxA5n8ikch3ej3st47j3AX8j+M41/dattmv70kkErkxEdRs6328A9TALu8Tx3G+jBt2nhGJRB7utf2PwDvALxL3gxuILI9EIuftMu632IcBvk7n4/5H7KmRSKSq174/2mXYPwIVu9aeON723wVe4PdAM7AgEolUJrb/Afd9/S3Hcf4ViURW7es5AK9FIpGenxnHcZbhvq8/hBtO7um5D9Z7fgbg4M70/Fk/6hcREemhGU8iInLIiEQiz+P+r/91QD7wMdwPlMscx3kqcca77T6AGyD8cJfQaftY8cTVCxKX/937LGWJ5Xz3Asc5jlO6y8Pv3zXoAi7Cncnxd8dxSnp/Jcbx0GtmxT48uj10StRige0fFi/otb0nlHAcJ9NxnGLc2Q6P4C7vmdbH2L/uI3TqGctxHOM4Tl6i7mrc2VZH9aPmDycuv71rL6BIJGK3v7aJ622JY3kdxylIHOvxxO69j/W+xOVOs8Uikchdibr6fA6JsbMSr0dWYuzpjuPk9eN5gLusqwx3ts3Gfj6mPz6Mu+Tsuj7eI/cAubgzgoCkfE8ORF/vk0txg9xXd6k9gDub5jjHcYKJfRuBUY7jHLcfxx7I69SYuLwwMatnN47jFAFnAA/tGjrBTr8L5uEuWf3H9tApcX8n7s+fBzfo6o9f7XJ7+/t7X7MH35e4TPZ7fvvrdJLjOGX7Kl5ERKQ3zXgSEZFDSqLn0kehZznUCcAncWcl3e04zrzEB8XtH/Be72ucXibgLmfaNUgC96x570vs03vZ0so+9p2euHxsL8casY9atuurlmWJy55wzXGcHOAHuDO2xvTxmMI+tvVVO4klhT/CnbGSvcvda/darWsKbljQZ/+tXY4Vwl2+dQTg3+Xu3jVPxP3e9FXzctwZHL3HLcOdFXc+bnC0qwLcpVX70oY7M+U23OV/J0Uikd0+9O+H6bhNx1fsZZ+e90gSvicHYk/v8SB7X8JXghvWfQd3meXTjuNU4i5vux+4PfHzuTcDeZ1+j/v9/iPwv47jPIO7HPCWSCSyvc7JifH687sA3J/7XW3fNrGP+/qy08zGSCRS6zgOuEsA92ZQ3vORSGR9Yqngt3GXTb6Buwz5tkgk8vK+noyIiBzaFDyJiMghKxKJrAeudxznBtx+PMfiNgR+ZpAP3drHNpO4/Ag7+ibtqj/L7AbiZtxGyX/BbYZei9tc+CzcpVh9zYzerXbHccYmHh/FDToiuA3WLfBr3J5V/WETX3uU6M91K/ASbk+ajbgNqr24gcF+zeZ2HMfgzvSaDvwGtz9OI+7r8THcIKnfY0cikXscx7kIt2fV9vBpb0FIfxjc1+fMRF19eReS+j3Zm739Hbmn9/jbuL3L9qQaepYcTsJtIn9S4utDwPccxzkuEons7Yx//X6dEoHOfNzg+VTcnme/An7oOM5ZiVmSQy4SieypbrOH7QM20Pd8JBL5nuM4/wDOxn29Pgl83XGcn0UikW8mqy4RETn4KHgSEZFDXiQSsY7jvIgbPI1KbN4+Y+Bw9jDLJ2EN7oez6bhng+ptRuKyP7NLtvd9qYlEInub9dQf0/vYtr2WNQCO4xTghk43RCKRq3rv6Lhn1huIC3CDjPMikcgTu4xVjNuXaV9W4gYFc3BDpT25DDdoOikSifSEG457hr5dbf/eTGX3WSi7vkazE8e+JhKJfL/3HY7jfLIf9e8mEonc67hn6bsDeMJxnJP7WGI5EKtwl3xt6Mc4A/2e7C3wq8Ndgrmr/s7e2W4VbuP4x3stT9ujSCTSjPva3QHgOM5ngD/gNpLfW7P9gbxO20OepYkvHMeZjduf7Xu4Ict7uK/P4fsYanswPLOP+2bsss9gGdT3fKLH3O+A3znu2fMeBr7hOM4vd+mRJSIi0kM9nkRE5JDhOM6pffVxSfSVOS1xc/uStNuBTtxT2u/W2ycxWwDc5UAA3+61DcdxDsM9u9wzvZbs7E0YNwz4Ya8+N72Pl++4p6Lvj1MTDZZ71/qNXerdPqNipxkUjtvkfKBBy57GugL3tOv9sb2J8U+c3c9M1/v1juGGAJ5d7uvrdPF3Jy6/vstY72OXJUfs+TkcRq++WAMViUTuA96PuwTwCcdxZuzjIXtzQ+LyJ4lG1jtxHKf3UsyBfk+a6TtcAjcUXOg4zvYzC+I4TiHurJiBuD5x7D5nPPWuP9GPaVfb+5btqc7t+v067eE4K3CXSxYBJGZXPQic2Vco2+u9+RqwAfiY4zjlve73474HLTvek4NlUN7zid8/Oy1rTfRi2x7s9bUsV0REBNCMJxERObT8Cih2HOce3CU/rbi9jT6EO0Pg+kQPKCKRyCbHcb6EO8Pibcc9G9h63BlR5wMfB96IRCKPOu4Z7D4IFDqOcx/uh+vP4s7M+UJ/Cksc79O4Z2Zbnlj+tx53hsgs3F5RM3BPL78vbwKPO+7ZtLYk6j0Fd3bT84njNTmO8whwqeM4bbhn9xsHfAp3hta+esn09iDua3mD455Brx539thZwGr68fdGJBJ5yXGc/wW+CbzmOM6twFbcvjkX4S6BbMANBC9MPL/rcXs8vQ+3IfKuYz7sOM69wOWJBtEPAZMSz/Ed3FPUb7ccd4bINxIBSwT3PfEp3PdKX6ei75dIJHJ/YongneyY+dRXH6B9jfOy4zg/wO3L9YbjOLcBlcDIRH1n4TbqhoF/T14APuE4zo9wX4s4cG+i+fTvgRtxX/MbcPv+XIH7/uxvsAjucq5TgZ87jnMybgPrKG5D7sUkZrIl9l3uOM4LwIu9nuOVuGHwv/d2kAG+Tn913DNYPpJ4PkHgYtwG5Nf3GvZzwHPAg47jXIc7IyqI26R9He7Z3mKO43wO9+xxLzuO8xfcBvMXA0cDP+nnGe322yC+508C/uI4zh2J/ZoT938SeDFJPcxEROQgpRlPIiJyKPkK7of/o3E/lP4FNxiqxF2+s9MMjkgk8ifcJTsrE/v9Hrgc90Nn77OVfRj3NO/jgF/inqb+SWBhJBLZV0Pi3sf7J26PmddxP/z9Efg87gfmq3GDmP64B/g0cG6inmm4fX4+sct+lwL/SOz3e9wA57u4YVu/RSKR1bjL5NbiNoX+Ke5skROATQMY51u4IWAj7gytX+POFnqARM+gSCTyb9wAIgf4RWK/CG4voL5cDPwfbnD1S9zeNO/H/R72PnYMd1nVvbjf498k6r8cuK+/z2Evz+0B3Fkk+bjh02H7eMiexvkh7hLJSuBLuN+rK3HPiPiFXvsN9HvyXdzA5LO4IdMtuKEnkUjkJtzXeSTua3kpcA1w7QBr78J9jb+YGPuHuGHwxbhLxP6n1+6/xH2tvgD8CbgKdwnmwsQZI/d1rH69Trizo7bgfp9/hztzrgu4KPHzv328tcCRwN9xZ0f+BjckLcQNrbbvdy9uiLYCd9bRT4FM4JORSOS7+6o7SQbjPf8m7u/OE3G/979JXP8Je/7ZExERAcBYu9ceniIiIjJMOI4zHjdo+GEkEvlBaqsREREREdGMJxERERERERERGSQKnkREREREREREZFAoeBIRERERERERkUGhHk8iIiIiIiIiIjIo9nl644OMraysTHUNSVFaWkp1dXVajpfOtWk8jZfK8dK5No2n8VI1lsbTeKkcL51r03gaL5XjpXNtGk/jpWqs4TBeKlVUVJg93aeldsOU3+9P2/HSuTaNp/FSOV4616bxNF6qxtJ4Gi+V46VzbRpP46VyvHSuTeNpvFSNNRzGS1cKnkREREREREREZFAoeBIRERERERERkUGh4ElERERERERERAaFgicRERERERERERkUaX1Wu1Ao9H3gB8CscDj8TigUOhr4MxAE1gGXhsPhqtRVKCIiIiIiIiIie5K2M55CodBc4GhgfeK2B7gR+Gw4HJ4KPAX8NHUVioiIiIiIiIjI3qRl8BQKhTKAPwCf7rV5HtAeDoefSdy+FggNdW0iIiIiIiIiItI/aRk8AdcAN4bD4XW9to0lMfsJIBwO1wCeUChUNMS1iYiIiIiIiIhIPxhrbapr2EkoFFoI/Bg4JRwO21AotA44B3CAj4fD4bN77dsKjA6Hw3X9HD69nqyIiIiIiIiIyPBn9nRHOjYXPwGYDqwNhUIAo4GHgd8C47bvFAqFSoD4AEInACorK5NXaQpVVFQk9bkkc7x0rk3jabxUjpfOtWk8jZeqsTSexkvleOlcm8bTeKkcL51r03gaL1VjDYfxUqmiomKP96Vd8BQOh39Kr6bhvWY8LQOuDIVCxyX6PF0F3JaSIkVEREREREREZJ/StcfTbsLhcBy4DPhTKBRahTsz6luprUpERERERERERPYk7WY87SocDo/vdf05YFbqqhERERERERERkf4aNjOeRERERERERERkeFHwJCIiIiIiIiIig0LBk4iIiIiIiIiIDAoFTyIiIiIiIiIiMijSvrm4iIiIiIiIiIi1Frq7ob0N2luho929nri07W3Q0dazrSEjg7jxQF4hJq8A8gsgrxBy8zFeb6qfziFDwZOIiIiIiIiIpIzt7oLlb2GXvU5tvJtYXd2OAKlXsERHG8Ri/RvU46HZH8B2tLvH6H2fMZCTB3kFkFeAyS90A6lEOGXyCiE/cTs7F+PZ92IxNxTrgs5O6OyArg73sqMDurZv68T2ut02czZUTBjgqzX8KHgSERERERERkSFluzrh3dexrz6HffMlaGuBQAYdhSXg80FmEILZUFiCyQxCRqa7LSMTMrMgM4jJzISMoLt9p+tB8PkZNWoUm9euhWg9RBsg2oBtrO+5bRsbIFqPfW85NNa7gRC7hFQeT09AVVVYTKy5aUeQtD1kSoRKWLv7E92L6LTZ8NUfJ+kVTV8KnkRERERERESGgLUW4nGIdbszd/px2dnZio2BCWaluvwDZjs64J1Xsa8+i33rFXcGU1YO5oijMfOOgemHUzFuHJWVlUk7psnIgNJy9wswe6rNWndWVbTBDaGi9YlgqiFxvd4NywIZkJOH8Qfc64HEZc/tHduMP2Ov+5RNmMSWbduS9lzTlYInERERERERSTobj0NDLR1NddiaWvAYd4mT8ey43G2bcWeY9N7m2fkx8eYmbGtL0uqMNTZga6t3zFzp7HCXdnV2YHtv6+PLduy+bYuNE+vo2HOgNEA9sUQwCwpLoKgEU1S643phCRS7t40/kLTXJVlseyv2rVewrz4H77zqvk45eZgFizDzjgVnFsaX+mjCGOO+xsEsGFHhbttlnxEVFckNxQ6RPlOp/+6KiIiIiIjIgNmODuwbL8C7r1FfOoJ4ILgjiCgsgYKiQf9ga2MxqK2C6q3Yqi1QtQVb7V5SvRW6u6hK8jE3J3m8AcUIxkAgc8cMlu1fGRmQkwuBEgJ5+cQ6Ot3lYl4veLdf9r6+50vT+7bHS2GGn7o1q6CuBltXA/U12PWroakR2GVZWG4+JEIpU+QGUxSVuu+JohLIL0riK7dntrUZ++bL2FefhXdfd3sf5RdijlnszmyaMvOQCV1EwZOIiIiIiMiwYeNxiLyNff4J7GvPu0uVcvJoiXVj21rdfbbvbDxug+TC4kQgVZq4LE7Mlil1Gyl79h4A2K4uqNmWCJUqE5db3XCptmrnZs+BAJSOhPJRmNlHQulIiidNoa6uzu1/Y+MQtz3Xe5ae2R3b2OM293p+bi6N0WjSXtP8wkIa2zvcpVG9w6Rdg6VAJvh87syYvShO8qyYrIoKGqbM2m277eqE+ppEIFUN20Opuhqo3oKNvAW7vic8HjYXFBHPyXeDoLyCRBPtQkxBosF24sxvJjM4oDptcxT7xovuzKblb7qzuwpLMCec4c5smuTs870mBycFTyIiIiIiImnObt6AfeEJ7ItPumFDMAsz/zjM0SfBlBlUjBpF5epVUF/rBhH11b2u18Dm9di33WVOuzVOLihKzJBxZ8pER5QTX7+mZwYT9TU7N00OZkHpSMzYSXDkcVA2ElM6EsrKIb9ot2AmWFGB2UMQs/cIp2+5FRU0JTHYSfZ4Q8X4A1BWAWUVe+5b1NaaCKSq3UCqrppgdyctlZvc5tob10JTA8Ri7NYWOyMzcZa3ot3O9GbyC93twSDNbz5PbMkDEHnbDQxLRmAWn+vObBo/pV9nhJODm4InERERERGRNGSj9diXnsI+/wRsWOOGRDPnYj7wMcycBe7snARjDCYrB7JyYNS4PoMIay20Nu88M6Z+x3W7/j1440UauzrdU82XjcRMnek2Ze4JlyogJ3efs34kPZhgFowaC6PG9rwniioqaO8VtNl4HFqadm6o3VgH28/41lgPlRuxy9+ERG+t3iFVPbjh1+nvd2c2jZ2o94fsRMGTiIiIiMh+sl1dsGktdk2ERq/BLjpTfUsOMrarCzrb3V5GQ3G8RN8m+8JSWPa6O4Nk3GTMB6/AzF/kLo3aT8YYyM51v8ZM2GM4NbK4mK11dft9HBlejMfj9obKzQfG73UWmu3q3HHWt8Z6bHOUsqOOo9ofVNgke6TgSURERESkH6y1ULMNuyYCa1e6lxvXQLd7lqooYDx+zPGnp7ZQwcZixJoasbVVbo+b9jZob8W2uZe0t7nbO9p67rftbTvt23OZ+P5u8nqhuAxKyjGlI9ylZqUjoMQ9TfuBnOrexuOw8p1E36bn3GMXlbgzSBaehBk5JlkvzT4ZY/BkZg7Z8WR4Mf6A+3NQXObeBgJ7WUopAgqeRERERET6ZNtaYd0q7JoIdu1KWBPpOYsUgYA7C2XxuZgJDkyYiu+fv6bz3luwR52IycjY++CyX6y10ByFhjpoqMM21O643ljXc51oA5U2vu8BAxmQGYTMxCnUM4NQXOo2VQ5mudszg5CRQY6N07xmFbZ6q7skraVp5544OXluAFUywl2aVlqOKS13g6nCoj6bKtvKXn2b6mogM4iZdwxm4cnuWb/UG0dEDgIKnkREREQk7dn2NuwDt1Ef8BOPA9luLxuTuNx+m6wcjN8/8PHjMajc4M5iWpOYzbR1046GyuWjMYfNg4kOZuJUqBiH8e38p3T+xz5H9TevxD5xH+aMC5PwrA8d1lrirc3YLRsTgdL2EKnWvd47VIp17z5ATh4UFENBEWbMBMgvJH/UGBo7Ot2ZSBm7BEnBIGQEB7QssqCigtbefXFam6F6G9Rsdc/wVr0NW70Fu24VvPosxOM7gimfD4pHQOkITEk55OSxdcWbxN9bvqNv04Ufxcw5SqGliBx0FDyJiIiISNqzN1+LfWEpLZlZ2LaWHdv72jkQ6AmhegKqXcIpsrNpLSkl/uarbsi0/j3oaHcfn5MLExzMgkXubKbxU9yAax8yD5sLh83DPng7dtHp/XqMuMvM4r/7EZvfeXX3O4NZOwKlKTPds68VFGES2ygock/73kfYONhnKjNZOTAuB8ZN2q0njo3FoK4aqrdia3aEUlRvw66OQFsLTJqGufgTmAXHu2cLExE5SCl4EhEREZG0Fn/ucezzT2DO/SCjr/oamzdudM+s1NrsfrU0u7NPWpp339baArVV2I1roKXF7emTUAvg9blNlo89BSZMdWczlY7c7ya5nvd/hPg1X8Q+fAfm/Zcn5wU4yNkXlsI7r5JzToiW8jE7QqX8QnfJ2zBkvN4dy+36uN92dFA+YQKV6osjIocABU8iIiIikrbs1k3Ym6+FqYdhzrkYSHyoz81zvxL6GxPZ7m53tklLM6UF+VR7A26z3CQxYyZgFpyAXXIv9uRz3BBF9si2t2LvvB4mTKXgU1+jbevWVJc0JLScTkQOJepWJyIiIiJpyXZ1Ev/Lz8Hvx/PJr/bZnHmgjM+Hyc3HlI8iMHFqUkOnnmO878MQi2HvvTXpYx9s7AO3Q2Mdng9eoUbaIiIHKf12FxEREZG0ZG/7J2xci+djX8IUDp+ZQ6a0HHP86dhnHsFu01KqPbFVW7CP3oVZeBJmopPqckREZJAoeBIRERGRtGNffwH7xP2YU87HzJ6f6nIGzJxzMfgD2LtuTHUpaSt+2z/B68O8/yOpLkVERAaRgicRERERSSu2tpr4v34L4yZjLhyeoYTJK8Scch72lWew699LdTlpxy57A954AXN2SH2wREQOcgqeRERERCRt2FiM+F9/DvEYniu/jvH5U13SfjOnXQA5ucTvvD7VpaQVG4sR//df3TO+nXJeqssREZFBpuBJRERERNKGvedmWL0Cc9lnMWUjU13OATFZ2ZgzPwDL3sAufzPV5aQNu/RB2LIRT+jjg9LcXURE0ouCJxERERFJC3bZG9gHb8ccdyqeBcenupykMCedBUUlxO+8HmttqstJOdsUxd5zE0yfA3OOSnU5IiIyBBQ8iYiIiEjK2Wg98b//H5SPxnzwylSXkzTGH8Cc9yFYtwpeez7V5aScvecmaG/Dc/EVGGNSXY6IiAwBBU8iIiIiklI2Hif+919DWyueT30Dk5GR6pKSyiw8CUaOIX7XDdhYLNXlpIzdtBb75MOYE8/CjBqb6nJERGSIKHgSERERkZSyD/8Hlr2O+eAnMaPGpbqcpDMeL54LLoOtm7HPLUl1OSlhrSX+779BdjbmvEtSXY6IiAwhBU8iIiIikjJ29QrsXTdg5h2LWXR6qssZPIcfBRMd7D23YDs7Ul3N0HvteYi8jTn/w5js3FRXIyIiQ0jBk4iIiIikhG1pJv7XX0BhCeYjnzuoe/4YY/C8/3JoqMU+cX+qyxlStrOD+G3/gFHjDu5wUURE+qTgSURERESGnLWW+PW/h4ZaPFd+HZOVneqSBp1xDoPD5mIfuB3b2pzqcoaMfeQuqK3C88ErMF5vqssREZEhpuBJRERERIacffIheO05zAWXYSY6qS5nyHgu+Ai0NmMfujPVpQwJW1eDffB2mHsMZtrsVJcjIiIpoOBJRERERIaU3bQWe+vf4LC5mFPfl+pyhpQZOxGz4HjsknuwDXWpLmfQ2Tuug3gczwc+lupSREQkRRQ8iYiIiMiQsR3txP/8c8jOxfOxL2E8h96fo+b8D0Mshr3v36kuZVDZ95ZhX3oSc/oFmJIRqS5HRERS5ND7l15ERESkH2xrC/EH7yB6279SXcpBxd7yZ9i2Gc8nvozJK0h1OSlhykZiFp2OffoR7LbKVJczKGw8Tvzff4OCYsyZF6W6HBERSSFfqgsQERERSSe2OYpdci92yX3Q1kIjYLwBPEefmOrShr34C0uxzy7BnB3CTJ+T6nJSypxzMfa5Jdi7b8Jc+fVUl5N09rklsP49zCe/isnITHU5IiKSQprxJCIiIgLYxnrit/+T+Lc+ib3vVpg+G893fknGzCOwN/4JW7Ul1SUOa3ZbJfbGP8HkGZhzL0l1OSln8gsxp5yPfflp7PrVqS4nqWxrC/bO62HSNMyC41NdjoiIpJiCJxERkTRircW+8xrdNdtSXcohw9ZVE7/lL8S/fQX2kbsxc47C84Pf4/30tzETplD0tR+B10v8Lz/HdnelutxhyXZ1Ef/Lz8Hnw3PFVzFeb6pLSgvm9AsgO5f4ndenupSksveHoTmK55IrMcakuhwREUkxLbUTERFJJyveIv6bH7DF48XMOwZzynmH1Knmh5Kt3op98Hbsc48DFrPwZMyZF2LKKnbaz1dWjufyzxP/0/9g/3MjRmfnGjB7x79gw2o8n/0upqg01eWkDZOVjTnrIuxt/8Quf/OgWH5ot27GLrkXc8xizLjJqS5HRETSgIInERGRNBK/799QUETuSWfR9NCd2JefdperLD4PM3ehZookgd2yEfvA7diXngSPF3P8aZjT348pLtvjY8zchZgTz8Q+8h/s9DmYw+YOXcHDXNsLT7pBxOJzMYcflepy0o456WzsY/cS/88NeKbNHvYzhOLhv4Pfj3n/ZakuRURE0oSCJxERkTRhV74DK9/FfPAKCi77FC0nn4N99nHsknuwf/kZtqgEc/I5mONOw2TnpLrcYcduWIN94Dbsa8+BP+DOJjv1fZiCon493nzg49hVy4j/41d4vv9bTH7hIFc8/Nm6aup+fQ2MnYi58KOpLictGX8Ac94l2Ot+B68/D3OPSXVJ+82+/Sq8/Qrmoo9h8vTzISIiLgVPIiIiaSJ+362QV4BZdBoAJjMLs/gc7ElnwtuvEn/0buzt/8Le+2/MMSdjTj4XUz4qxVUfGNvWSsvSh7BdMSgqhaJSTEZGco+xJkL8/jC89TIEszBnfsANnXLzBjSOCWTgueLrxH/yFeL/+DWeL34f41G7zL7YpiisfIf4Q3dgurvwXPkNjN+f6rLSlll4MvaRu9xZT3OOSvrMRltbhX3+CeqJY2cvwIydlNTxAWx3F/Hw36CsArP4nKSPLyIiw5eCJxERkTRgV6+A5W+6MwUCOwcvxuOFOQvwzlmA3bgW+9g92KcfwT7xAMw6Es+p58MwXKJju7uJ/+G/qYu8vfMdOXk7QqjinS8pLoXcgn0+V2strHyX+P23wvI3ITsXc/6HMSefjcna/9liZtRYzMWfxN7wR+yjd2FOf/9+j3UwsS3NsOod7Iq3sZG3YdM6946MIEVf/j4NIyr2+vhDnfF68bzvUreP2HNLesLnA2Hb27CvPef2MEv8jDX7A3D3LTBuMuaEMzDzF2Eygwd8LMD9fbR1M57PX43xKWQUEZEdFDyJiIikgfj9YcjJxZxwxl73M2MmYD72ReyFH8EufQi79AHi/3c1jBrn9tA5+kSMPzBEVR8Ye/s/IfI2hZ/5Fo3Z+di6KqithroabF01bNuMXf4GdLRjez/Q54eikp0DqaJSt2l1cRltleuI33AtvLfMnUF20cfcD9lJ+oBtFp2OffcN7H9uwE49DDNhalLGHU5sWyusehcbeRu74m3YuAasBX8AJk/HvO9SjDMLxk8ha+xYGiorU11y+jviaJgwFXvPLdijTtgtgO4PG49D5G3s849jX3seOtqhtBxz/ocwR5/EyAmTqLzr39inHsJe/3ts+O/u74zjz8CMmbDfpdtoA/beW+CwuTDryP0eR0REDk4KnkRERFLMrn/P7Yvyvkv7HY6YvEK3L8yZF2Ffegr72N3uB8n/3OCGLCec2e/eRakQf3aJ23D6lPPJOfsiopWVGGbutp+1FlpboK4aaqvcQKpuRzhl330dGuvB2p5wqgagsARzyZWY407drw/we2OMwfORzxG/ZhXxv/4Cz9W/xgSzknqMdGPb2+C95W7QFHkb1r8H8Tj4fDBxGuacD2KmzYIJjpbU7SdjDJ4LLyf+i+9in3gAc/oF/X6s3Vbphk3PP+H+fASzMAuOxxxzMkya3jND0JObh2fxOdiTz4bVy7FPPox95jHs0gdhwlTM8ae7s6AyMgdUu73rRujswBP65LCbeSkiIoNPwZOIiEiKxe8LQ1Y25uSB90Uxfj/m2MXYY06GyNvEH7sHe38Y++AdmAWL3F5Gg9DP5UDYtSuxN/7RXR540Uf3uq8xBrJz3K8xE+jrI63t7oL6WjeMqq2iqLSU+gnTBnW5j8nOwXPFV4n//DvYG/8En/zKQfWB23Z2wOoViaVzb8G6VRCLgdfrBhRnXuTOaJo0LenB3qHMOLPgsLluE/xFp+51X9vajH3lGXcp3eoVYDwwYw7mwssxhx+11++LMQYmz8BMnoH94Cexzz+Bfeph7HW/23kW1Ojx+6zZrl+NfeZR98ybI0cP9CmLiMghQMGTiIhICtlNa+GNFzDnfvCAZs0YY2DabLzTZruzHx6/D/vsY+4MiKmH0f2NHwOpb4RtG+uJ//F/IL/QbTidhCbKxueH0nJ3SRGQVVExJEu7zOQZmHMvwd59E8w8HHPM4kE/5mCyXZ3Yx++nasWbxFe8Bd3d4PG4/YBOex/Gme0uoxvgbBgZGM8FlxH/0ZexD/8HJn9zp/tsPAbL3sA+9zj2jRehqxNGjnHDpqNPxBQUD/h4JjsXc8p52MXnwqpl7jK8px91ezZNdNwA6sjj+mz6b60l/u+/Qk4e5tyL9/s5i4jIwU3Bk4iISArZ+2+DzCBm8XlJG9OMqMBcciX2/A9hn3kUe/9tVH37KuyXr8GUliftOANlu7uIX/u/0NqE51s/H/BZ5dKROesi7Iq3sDf/GTtx2rA9y6Bd9jrxm66Fqi3EJ01zz5g4bZY7K+YgX0aYbszYSZgFx2Mfu4fYJZ8AwFZucMOmF5ZCY53bLP+4UzALF8P4yUmZbWeMgakzMVNnYj94RWIW1EPYf/0Ge+vfMAtPcpfijRrX85i2px+F95ZhLvvsATXtFxGRg5uCJxERkRSxWzZiX30Wc8aFmOzkf2gzWTmY0y7ATpuN/dX3if/ye3i+9t+YkhFJP1Z/2Fv/5n5IveJrB9TIOJ0YjxfPJ75C/JovEP/rz91AbRj1OLKN9djw37EvPQVlFXi+fA3lp5xFpZqBp5Q5/0PYV5+l9hdXE2uod3tqeb1w2Dw8x5wMs+YP6vvM5ORhTj0fe8p5bhP5Jx92Q6jH73OXVx5/Bmb2kTT84zcwdiLmuFMGrRYRERn+FDyJiIikiH3gNvAHMKeeP6jHMWMnUfLff2Tbt68i/ovv4vn6TzDFZYN6zF3Fn3oYu/RBzOnvx7Pg+CE99mAzhcV4PvpF4r//MfbO6zAXfzLVJe2Tjcfcnj533gBdHe5SzzMvGjZnRDzYmbIKzPFn0PHE/W5vs4s/gVlwAiavYGjrMAamHoaZehi26Qrs80uwTz2C/eevsV4fxLrxfP1/MJ4DXzIrIiIHLwVPIiIiKWCrKrEvPoU59TxMbv6gHy8weRqer/yI+P99j/jPvzOk4ZNdvQJ7859hxhGY9182JMccambOAszJ52Afuwc7fQ5m9vxUl7RHdsNq4jf+CdauhOlz8HzoqmG7RPBgZkIfp/zDV1BlU9+bDcDk5rkzKE99H6x8B/vUI+SMn0jr1N3PRikiItJbevxLJiIicoixD9wOPh/mtP6fMv1AmXGT8HzlR9DW4p6yvbZ60I9pG2qJ/+mnUFSC58qvHdQzI8xFH4XRE4j/8zfYhtpUl7Mb295K/Na/E//xV6FmG+aTX8Xz5WsUOqUp4/PjS8OzxBljMM4sPFd8lYLLP5vqckREZBhQ8CQiIjLEbM027AtPYBadhskvHNJjm3GT8Xz5GmhpJv7L72LrBi98sl1dbujU3ornM9/BZOcO2rHSgfEH8Fz5dejsIP73X7lnIEsD1lrsa88Rv/qz2CX3YI4/Dc+P/oTnqBOS0pRaREREZG8UPImIiAwx+9AdYAzm9Pen5Phm/BQ3fGqOujOf6mqSfgxrLfbma2FNBM/HvoQZPT7px0hHZuRozCVXwoq3sA/ekepysDXbiP/uR24AmJOH55v/i+fSzwxKM3sRERGRvih4EhERGUK2vhb77GOYY07BFJWkrA4zoXf49J2kh0/2yQexzzyKOSuEmXdMUsdOd+bYUzDzF2HvuRm7ekVKarDd3cQfvIP49z8LK9/BhD6B53v/h5k0LSX1iIiIyKFLwZOIiMgQsg/fCdZizrww1aVgJkzF86UfQlOju+yuPjl9iezKd7H//ivMOhJz/iVJGXM4McZgLv0MFJUS/+svsK3NQ3p8u2oZ8R99CXvndXDYPDzX/AHPqedjvAdvfy0RERFJXwqeREREhohtrMc+9TDm6BMxJSNSXQ4AZqLjhk/RBnfZ3QGGT7auhvi1P4XiEXg++ZWDupn43pisbDxXfA0aarHX/wFr7aAf0zZHiV/3O+I/+xZ0tOP53NV4P/1tTFHpoB9bREREZE8UPImIiAwR+8h/oLsbc9YHUl3KTsykaW741FhP/Jff2+8zstmuTuJ//Al0deL53HcxWYd2HyEz0cGc/2Hsq89in3l00I5jrSX+7BLiV38a+/zjmDMuxPPD32PmzB+0Y4qIiIj0l4InERGRIWCbGrFLH8QsWIQpq0h1Obtxw6fvQ0Md8V98D9tQN6DHW2uxN/wR1r+H5xNfxowcM0iVDi/m9PfD9DnYf/8FW7khqWPb1hbsqmVUf+tT2H/9BspH47n613guvByTkZnUY4mIiIjsL1+qCxARETkU2Efvhq5OzNmhVJeyR2byDDxf/D7x3/yA+C+/i+drP8HkF/brsfbx+9zZNudegjn86EGudPgwHg+ej3+Z+DVfJP6Xn+P57i8H9Hjb1gpVldiqLbCtcsf1qi3Q1AhAV24+5iOfc5uae/R/iiIiIpJeFDyJiIgMMtvShH3ifszcY9J+JpCZMgPPF75P/Lc/JP6L7+L52n/vM3yykbex4b/D4Udhzrl4iCodPkxBEZ6PfYn4b3+Ive0f8LVrdrrftrdC1Rbsti1QVQnbKrHViaApES71KCiGERWYw49yL8sqKD9+MduaWobwGYmIiIj0n4InERGRQWaX3AvtbZhz0ne2U29m6kw8X/gv4r+9hvgvv4fnaz/G5PUdPtnaKuLX/i+UVeD5+Jc142YPzKx5mFPPxz56N/V5+cRrqnYETdGGnXcuKIKyCsycBe7liJFQVgGlIzEZGbuN7c3NBwVPIiIikqYUPImIiAwi29bqBk+HH40ZPSHV5fSbmXoYns//V2LmU9/hk+3ocJuJx7rxfPa7mGBWiqodHswFH8GuWkbz3bdAfiGUjcTMOrJn5hJlI91t6s8kIiIiBxEFTyIiIoPIPn4ftLbgGSaznXozzmG9Zj5djeerP8bkFQCJZuLX/x42rsXzue9hykeltthhwPj9eL75U0aWlrK1viHV5YiIiIgMCc2HFxERGSS2vQ372N0w60jMuMmpLme/GGcWns9fDTVbif/f1dhEzyH76F3Yl57EnP9hzOz5Ka5y+DA+Px7NDBMREZFDiIInERGRQWKffAiam/Ck8Zns+sNMm43nc1dD9Rbiv/weLUsfwt5+Hcw9BnPWB1JdnoiIiIikMQVPIiIig8B2dmAf+Q9Mn4OZNC3V5RwwM32OGz5VbaHu59+DijF4PvZFjDGpLk1ERERE0piCJxERkUFgn34Eog14zrk41aUkjZk+B8/nryZjzgI8n/k2JjOY6pJEREREJM2pubiIiEiS2a4u7EN3wtSZmKmHpbqcpDLT51C2+EwqKytTXYqIiIiIDAOa8SQiIpJk9tnHoKEWz9kHz2wnEREREZH9oeBJREQkiWx3N/ahO2CiA9PnpLocEREREZGUUvAkIiKSRPaFJ6C2Cs85F6vxtoiIiIgc8hQ8iYiIJImNxbAP3AbjJsNh81JdjoiIiIhIyil4EhERSRL78lNQvRXP2SHNdhIRERERIU3PahcKhe4CJgBxoBn4fDgcfiMUCk0FrgOKgVrgI+FweFXKChUREUmwsRj2/ttg1DiYsyDV5YiIiIiIpIV0nfF0eTgcnhMOh48AfgH8I7H9WuAP4XB4KvAH4M+pKlBERKS3tmeXwNZNmLMvxnjS9Z9XEREREZGhlZZ/GYfD4cZeN/OBeCgUKgPmArcktt8CzA2FQqVDXZ+IiEhvNh4n+u+/Q/lozLyFqS5HRERERCRtGGttqmvoUygU+htwGmCAM4BM4PpwODyz1z7LgEvD4fBr/Rw2PZ+siIgMa63PPUHtf3+doq9eQ/bJZ6W6HBEROYhYa+mMxensjtPRHd/p+vbbXo9hzqh8POovKCKps8dfQGnZ4wkgHA5/EiAUCl0G/By4OhnjVlZWJmOYlKuoqEjqc0nmeOlcm8bTeKkcL51r03gDY7u6oKoStm7CbtmEfW4JvpGjaZgyi8ZD4P2SzrVpPI2XqrE0nsbrD2sty6vbWLKmkS7jJ9rSRlcsTmfM0hW37uX22zHbs70/Fk4o4jNzi8jLPPCPeOn42mk8jZfqsYbDeKlUUVGxx/vSNnjaLhwO3xAKhf4CbAJGhUIhbzgcjoVCIS9QAWxMbYUiInKwsi1NsGUTduumHZdbN0H1NrDxHTsWl1Fw1dep93pTV6yIiKStWNzy/MYm7lpex6radrL9HioKsiAeI+A1ZAW8BLym58vv8biX2297E7c9O9/evs+6+g6uf6OGL22L8vVjK5helpXqpywi0iPtgqdQKJQDFIbD4Y2J2+cCdUAV8AZwCXBj4vL1cDhcnaJSRUTkIGDjcaivgS0bEwHTZuzWjbBlEzT1ajno88OICsyYibDgeLef08jRMGIUJiOTYEUF9QfJ/1gNpfsidTx0/3pG5XqZUZrFjLIgEwoz8Xm0XEREhr/WrhiPvtfIfZE6qlq6GZnr51PzR3DyxHwmjh2dtJkOs0Zkc+LMcXztzjf4zmMb+MjhpbxvehFGS+9EJA2kXfAEZAO3hUKhbCCGGzqdGw6HbSgUugq4LhQK/RdQD3wkhXWKiPSLtRb70J1sevB2bFcnGJNYAW3c6yRu976+/b6d9kvsk9hva2ER8aNPwhy7GJN5aPzPprUWuruhqxO6u6CrK3HZ63av63bX+7bf7uqitq2Z2JqVsG0zdHbuOEh2LowcjZmzYEe4VD4aSsowHs1oSqbwOzXc9GYN00fksq6+nRc2NgOQ6TNMLQkyozTIjLIsphYHCfqH7nwo1lpq27rZ2NjJxsYO8rfEmJwdY1ReYMhqEJHhrbqli/si9TzyXgOtXXFmlAb5xLwRzB+Vg3eQgnVnRC7/d+Z4fv/iVv71ejXvVrXxxYUjyc3Qv10iklppFzyFw+FtwNF7uG8FcNTQViQisv9sRzv2X7/FvvIMmfOPo6OoDLCJUx1YsLtch8Q2u8/9TPUW7L//ir37Jsyxp2JOPhtTWj7kz3Gw2bUrid/6NzZuWOOGRsng89FRXAalIzHObBg5ClM+BkaOhpw8/Q/xILPWcuObNdz+bi0njM/jpxfOpWrrVmpbu1he3cay6jaWVbVy69u1WGrxGJhYmMmMsiAzSrOYXhakIAk9TKy11LR2s7Gxg42NnWxo7Oi53trVayklVQBU5AZYMDqHI0dlM700S7Oy9mFlTRv3RuqZMrKdI4oNY/IzUl2SyKB7r7adu1fU8ez6KBY4Zmwu508rYmpJcEiOnx3w8o3jKrh/ZT3/fK2Krzy4lq8dNwpniI4vItKXtAueREQOFrZ6K/E//DdUbsRc9FFKPvpZtmzZkrTxR1RUsPnZpdjH7sU+cR92yb1w+AI8i8+DqTOHfXhim6LY/1yPfeZRyCsg99yLae7sAr/fXfbmD4DPB/4Axh9IbOt1n98HvsAu1/3g9WE8noOqmeNwYq3l769Vce+Kek6dlM+nF5Tj87izmYqz/Bw3zs9x4/IAaOmMEalpY1lVG8urW3loVQP3rKgH3BDIDaLcWVHlOf49vufj1lLd0tUrXOrsCZjau3cETPmZXsbkZ3DihDzG5GcwJj/A2PwM8otLuf+1Nby0uZn7IvXctbyO7ICHuSOzmT8qh7kVOZpR0EtltJMb3qzmuQ1NZPk9PL3O/QA+viCDRePzWDQulxE5mj0mB4+4tbyyuZm7l9fxTlUbQZ+Hs51CznWKKMvxD3k9xhjOcYpwSoL87OnNfOfR9Xz0iDLOcQqH7d8G1lpaOuPUtXVT29bNuo5a8uPdFAb1cVZkONBPqojIILDL3iD+l5+DtXi++H3MzCMG5Y89M2Eq5oqvYi/6KHbpA9inHiL++gswZgLmlPMw84/H+If+j94DYeMx7FOPYO+6Edpa3Odx7iUUTJpMq4KiYS1uLde+tI2H32vgHKeQT84r2+vPRXbAy9wKN9gB6IrFWV3XwbKqVpZVt/HCxiYeW+324SrM9DK9LIsZpUGcFj9vr6vtCZc2RTto795xVqjCTC9jCjJYPCmfMXkBxhZkMCYvsMczQVXkBznbKeRsp5DWrhhvbmnl5c3NvFLZzNPrm/AYmF4a5MhROSwYlcOovMCw/XB3IBrauvn32zU88l4Dfq/hg7OKOX96EQXFZdzx0iqeXh/lhjequeGNapySTBaNy+PYcXkU6YOjDFMd3XGeWNvIPSvq2RztpCTLx8fmlnLqpAKyA6kPo6cUB/nVmRP4zQtb+NurVbxb1crnjh5JThrU1lt7d5za1m7q2rqoa+2mrs39qm3tpr5tx+3OWO+z+7nnlyoK+phUlMGkokwmFmUyqSiT4qDvkPwdLJLO9C+9iEgSWWuxj9yFveM6qBiD5zPfwZSNHPTjmsJizAWXYc8KYV9cin3sHuw/f4O9/V+YE8/CnHgGJq9w0Os4UHZNhPjNf4b174EzC88ln8KMGpvqsiQJYnHLb1/YwtK1US6aWcylc0oG/MHA7/UwrTTItNIg78cNsjY1dvJuVSvLq91ZUc9taIJX3aVxxUEfY/IDnDqpoCdcGp2fcUCzk7L8XhaOzWXh2Fzi1rKqtp2XN7kh1HWvV3Pd69WU5/iZPzqH+aNymFGahd97cH8AauuKc/fyOv6zvI7OWJzTJxdw8aySnpkIJTkZnDutiHOnFbGtuZOn1zfxzPoof3u1in+8VsVhZVksGp/HwjG5mjkmw0JDWzf3r6znoVUNRDtiTCrK5KvHVnDM2Ny0W4Kbk+HlO8eP4u4VdVz/ejVffXAd31g0iklFmUNWQ0N7NysiVazaXOeGSL3Cpbq27l2WNrsyvIbiLB9FQR9TS4IUBX07vrJ8FBUX8/Kqzayua2d1XTuvVrYQT+RS+ZleJhW6QdTkokwmFmVQlr3nWbGy/6Lt3ayp72BdRy3R+hZ8HrPjK3EWRl+fX+j7cYhR8CQikiS2owN7/e+wLz0F847B89EvYjKHtqeCycjAHH86dtFpsPwN4o/di733FuyDt2EWnIBZfC5m7MQhrak/bFMj9s7EsrqCIswVX8PMX6Q/Sg4SXTHL/z1XyXMbmvjw7BJCs0qSMq7HGMYWZDC2IIMzp7rBanVLF97sAgKd0UH/X32PMTglQZySIJceXkp1SxevbG7m5c3NPLSygXtX1JPl93BEYknevFE55B1EwUp33PLoew38++0aGtpjLByTy2WHl+61CfuInAAXzSzmopnFbGjs4Ol1UZ5eH+UPL27lzy9v5YiR2Swal8eC0blD2lBepD/W1LTw1xe28OTaKN1xy/zROZw/rYiZZcG0/vfKGMP7phczrSSLnz2zmW88vJ5PzCvjzCkFg1Z3a1eMFzc289S6KG9s3REK+TzuLKXCoJ+xBRkcPjJ7p0CpOHEZ9Hn2WltFRSEjvW09t9u746yr7+gJotbUt/OfZbVsnySVG/D0zIiaWJjJ5OJMRuT48aTx9y3ddHTHWVPfzqradlbWtLGqtp2tzV2JezcOeDyfhz5DqQyfhyPHNeEUwOwR2fq34CCh4ElEJAlsbZXbz2nTOswFl2HOvCilf4QaY2DGEXhnHIHdugn7+H3YZ5dgn1viziRafC7MmZ/ys7TZeAz75MPYu26AjnbMaRdgzr34kDlL36GgMxbnZ09v5uXNLXx8bhnnTy8a1OOVZvupqMinsrJlUI+zp2OfObWQM6cW0t4d580tLby0uZlXNzfz7AZ3SZ5TEmT2mGZiHW1keN0/sAOJy0yfIcPrIcPnIaPnunsZSFwO1tmwBsJay/Mbm7jhjRoqmzqZURrkOyeUDbh58dj8DD48p5QPzS5hdV0HT693Q6iXN28h4N3K/FE5LBqfx7yKbALeA//g0dEdJ9oRo6kjRjTxlVtnmJ4XJ9N38H6w6Y5bVte1s7y6lfYVLdjOVnICXrIDXrL9Hvcy4CHb714G/Z5h+2G8O27ZHO1kXX076xs64N0mmptbsFjiPefp6HXdQhzrXibu67luLZYd19u640Rq2gl4DYsn5XPetKJhd6bLaaVBfn3meH79/Bb+/PI23q1q5bNHlZPlT87fAl0xy+tb3LDpxU3NdMYsZdk+3j+jmHOOmIC3rYHcDO+g/H2U6dsxI3a7zlic9Q07wqjVdR3cs6Ke7kQKluX3MLHQXaY3v9FDmbdTM6MS4tayKdrJqpo2Vta2s6q2jXX1HT1BXnGWj6nFQU6fXMCk4kxGl5exdVs1XXFLdx9fXbHEZe/tsd770HNfc0eMB5dt5Y6uGD4PzCjN4oiKbOZV5DA2/9Bcyn4wUPAkInKA7Iq3iP/5ZxCL4fn8f2FmzUt1STsx5aMxH7oKe/6l2GcexT5+H/E//gRKRmAWn+OeES849EGPXb2C+M3XwoY1MG02nkuuxFRoWd3BpL07zk+e3MSbW1u5av6InllJh4JMn4ejxuRy1Bh3Sd7quna3L9TmZu57ZwttXbGeGQAD4f5v8M6h1MSyWibmGg4bkcW4goxBDQ3erWrluteriNS0MyY/wHdPGMX8UTkH9EHAGMPkYncGwuVHlLK8uo2n10V5bkMTzyYalB89Jpfjx+cxe4T7u2r7h5PGjhjRjm43SGrfOVTq+Wp37++I9fWCV1Ka5eOKI0dw1Jjc/X4O6aStK+425a9uZVlVG5Gatp7eOFl+L61dsb0+3mPcD+R7CqayA15yErcndGTgae+gNMs/5LMSGtq7WVffwbqG9sSl29Nte6jg80Buph8bj2OMwQMY4z4/YwyGvq97DBh2v+41hk8dO4Hjyr177Ac3HORl+vjeiaO5c1kdN71ZzZq6dr6xaBQTCvdv6V3cWpZXt/HUuijPro/S1BknN8PL4on5HD8+j2mlQTzGUDEyj8rK5iQ/m70LeD1MKQ4ypXhHGNUVs2xs7B1GtfPgqgbuTpy4oiDTi1MSZGpJkGklQSYXZ6YsmN5+5tVN0U7WttfQ3tTa83OZE/AS8JqkhTC1rV07zWRaVdtOW+LkG1l+D5OLM7lgRjFTE7+ri7N27h9aUZFPMcn7D5/SEeUseXM1r1W28NqWlp6l7MVZPuaOzGZuRTZzyrPTopea9M/w/a0pIgLY1hZY/gb27Vexq1dQN+dI7LxFMH7yoP+PiLUWu+Qe7G3/hBGj8Hz2u5gRFYN6zANhsnMwp1+APeU8eOMF4o/dg73179i7b8YcewpdF16K9WZgPIP7B5aNNmDvuM6dfVVQjLnyG5gjj9X/YB1kWrti/OiJTayoaeOLC0dy8sT8VJeUMh5jej78fGh2ac8ZFbvjlo7uOB2xxGWv652xXe6Lxeno3rGtvTtOZ7c7C2PFtiYeX9kOQHbAw4zSLGaWBZlZlsWkosykzJLa0NjB9a9X8/LmZoqCPj53VDknT8xP+gwsjzHMLMtiZlkWVxw5gre2tfLUuigvbGzi8TWNZPs9eDyraero3uMYWX4PeRlecjO8FGZ6GVcQIC/DR26Gl7zeX5lePFkF/PShZfzkqc3MH5XDlUeOSMlZyA5EQ1t3T8i0rLqNtfXtxK0bpEwozOD0yQVMLwsyvTSLwyaNZeOmzbR2xWnpjNGy/bIzTkuXe9ncGdtt25ZoF81d7bR0xnZq1A87ztSaG/BQku2nNNtPaZbPvUx8lWT5KAz69isU7YpZNkU7esKldfXtrGvooKF9R4BWFPQxviCDI0ZmM64ggwmFmYzKCzB29Kiknr30YDkbqscYLppZzPSSID9/tpJvPLyeK44cwamT8vv9b/G6+naeWhflqXVRqlu7yfAajhqdywkT8phTnp22ve38XsPERCPyUxPbYnFLqz+PZ5ZvIFLTRqSmnRc3uSGZx8C4ggymJcIopyRIRW5yZ0X1PvPq9pNibGjsYFNjZ0/409dSNp/H7BYKbw+l+gyLE/cH/R62bKjnuUgtq2rdGU21re7vVK+B8YWZnDghj6klQaYUuz9LQz0L0u/1MLs8m9nl2XwUqGnt4vXKFl6tbOHZDU08uroRj4FpJUHmJmZDjS8c3P94kQOj4ElEhhVrLWxeh337New7r8DqFRCLQVY2jJ9K69KHsA/f5Z7VbdHpmKNOwGRlJ7+Ozg7sjX/EPv8EHH40nk98adgsDzNeL8w7Fu+8Y7HrVmGX3Itd+gBbl9wLwWw3tBs/BTNhKoyfgiksTspxbSyGffJB7F03QWcH5owLMWeHhrwPlgy+po4YP3xiI2vq2vnqsRUcNy4v1SWlJZ/H4At4OdDfUBUVFby5aj3vVrXyblUr72xr4+XN7oem7ctPDksEUVOKM/EPYNlabWsXN79Vw+NrGsn0ebhsTinnTiskYwhmAHg9hiNGZnPEyGw+vWCE+z/flS0U5OXg6W7fOUTKcGeh5Aa8A/rAW1FRyP+dNZ57V9Tx77dr+Ox9a/jgrBLOm1aUlh+crbVsaerqCZqWV7dS2eT2WAl4DVNLglw0s5gZZVk4JZl9LqHyegy5iWBuf3THLa2dMZo74/hyCli+fgvVrd3UtHRR3dLFtqYu3t7a2usDs8vngZIsvxtO7RJMlWb5KMn2U9PcwWuVzTtCpoYONjXuWN7j9xjGFgSYW5HDhMIMxhdkMK4gg/xhPAMplWaOyOLXZ43nV89W8ocXt/LutlauWlC+x9lr1S1dPLUuypProqxv6MBj4IiR2Vx6eClHDeO+bF6PwRmRS26ssGdmbrQjxsqatkQQ1cbStVEeXNUAuCHr9hDKnR3V98/aruLWUtW8I2DasLczr+ZncHLizKtj8jMYVV7GusptNHfEegLhXcPj5s4Y1S1dtCR+Prv7Ma22PMfPzNIsppZkMqU4yMSijKQsbU62kiw/p04u4NTJBXTHLZGatsS/Cc3c+GYNN75ZQ0Gml7kV2RwxMocjRmbrZBVpRr+lRSTt2fZWWPYm9p1XsW+/Cg217h1jJmBOfz/msHkw0cF4vZQX5FF5163Ypx7C3nwt9vZ/YuYfh1l0urtPEv4nxNZVE//j/8D69zDnfwhzVmjQZwkNFjN+CuYTX8FeeDn5G1fT8MYr2HUrsY/8BxtL/G9yQZEbQI2fgpkwBcZNwWTnDOg49r1lxG/6M2xaC9PnuGerGzl6EJ6RpFpDezc/eHwjGxs7+ebxozhq9MGxfCndlWb7OXFCPidOcGeW1bd19wRR725r48Y3a4Ad4cTMsiCHlWXhlAT7DJFaOmPcuayOe1bUEbeWs51CQjOLU7bEKOB1l9sdPSY36bNOfB7DBTOKOW5cHn97dRvXv1HNE2sb+fT8cmaOSO1/KHTH46yqbWN5dVtP0LR9pk9uwMP0sixOm1zAjLIsJhZmDklY5vMY8jJ95GVCRUUBZZ7WPvdrSXwIrmntprqli6qWLmpauqlu7eLtba3UtXX3sdx0Zc+14iwfEwoymD8qJzGLKYOK3EBa9Dk7mBRk+vivk8Zw+7u13PJWDe/VtfPNRaOoSEzgjnbEeHa9O7NpWbXbzNspCXLlkSM4blzuQRv65WV4OXJUDkeOcv/eicXdnkeRXmHUq4leggYYkx/oCaKc0iCxYCuvbGpiY0NiFlPUDZk6ey35LQr6GJs48+qY/AzG5u/5zKsVI/MotANbqtgZi/eaxbhzUDVtbDnFpnVYnvDC59kxM/ayw0upb+vm9S1uCPXypmYeXxPFY2BKcSZzK3I4IRakobaVzkSvqc6YO6u4M7bjdtf22/Ed9/dsi8V7PdaSE9zC3BEBjh+fT1Hw4Hz/Dwa9UiKSdqy1ULkxETS9Au8th1g3BLNg+uGYWfMwh83FFOw+E8eTlYPnxDOxJ5wB69/DPvUw9qWnsM8ugVHjMMefjjn6REzWwIKTntpWvkP82v+Frk48n/seZs6CA326acEUFJMzYxbRWe7zsZ0dsHEtdt0qWLcKu3YV9o0X6flzqawCM34KTJiMGT8Vxk7EBDJ2G9dG67G3/8udGVZYgueqb8LcY7Ss7iBV29rFfy3ZSFVLF987cTRHjEz+bEPpn8Kgj+PG5fXMNot2xFi2PYiqauO2d2q51dbi88DkomDP0rwpxZksfWUjf3tuDU0dMY4fn8elc0oYkTO8mijvj9JsP98+fjQvb2rmL69s4zuPbeDkiXl89IiyIf1wHbeWt7e18th7jbxSuaqnJ1NZtp/DR2YzozSL6WVBRqdg+ctAZCcamI/fQ2u3WNxS2+oGUdWJUKqsuIAiTwfjCvr+8C2Dw+sxXDyrhOmlQX75bCVfe2gdl9fD6+ureX1LM91xGJ0X4MNzSjh+XB7luQf/74NdeT2GcYkZdqdNLgCguTPGqtp2IjVtrKxp4/mN7hIw19qex5Zk+RiTn8EZU7ISAVMGo/MDg37m1YDXQyDoobCPcKSiopjKyo5BPf5QKQz6OHliPidPzCcWt7xX186rlc28VtnCv9+q4Za3avo1jsH9z5mA1+D3ena77vcasv0eWrrj/PM1t+fU7BFZnDghn6PHDN8Zf0NFwZOI9Ju1+9EJt79jt7fBirfcXk3vvAp11e4do8ZhTjkPM+tImDQN4+vfry1jTM8sHRv6OPalp90Q6pa/YG//l9tT6PjTYdL0foUg1lrs0gewt/4NSsvxfOa7B/WMHRPIcF/vSdN6ttnWZlj3nrs8b90q7Mp34KUn3TDK43G/V+On9LzuTa88Rfz6P0JnJ+bMCzFnX4zJ2L/mpZL+qpq7uHrJBhraY/zgpDEpnykiO8vL8PbMGAJ3NsqK6jbeSQRRdy2v445ldT37zynP4vIjyphUdOj9zM4fncPs8izC79Ry1/JaXtrUzEcOL+PUyfmDGvRUt3SxZE0jS1Y3UtXSRU7Aw+nTy5mYCzPKgpRkDa/eU/vi9RjKcvw79dQ6WHooDVezy7P59VkT+MWzlfzlubUUB32c4xRxwvg8JhRm6D+NdpET8PYsBwY3NK5s6mRlTTt5+fnk2jbG5AeSdtZA2Tevx/TMPPvQ7FIa27upI5vG+jr824MkjyHg9ey47XVvexMnGNiXiooKXlqxlifXRlm6Nsqvn99CxktbOXpMLicmepxpZubuFDyJHMRsdzexuhrstkroaIP2duhoh442bEd74nabu629rWcf29F7+47H0NHORo8HApkQyICMjMRl4nYgww0WerYFeu2bCRkZbqCxfZvXS9OLTxB79nFY9S50d0NGEKbPcXv/HDYXU1R6wK+DycxyQ6bjT8euX419+mHsi0+6s3BGjnFnQS08CZPd95Ig29WJvela7LOPwez5eD7xlUHpG5XuTFYOzDgcM+Pwnm22oTYxIyoRSL36LDz9CBZoAJhxBJ5LrsCUH7whXbpraOvm8TWNTGvyUe7rHpRp4VuaOvneYxto645zzeIxOCXq25XusgNe5o3KYV5iGUl7d7xn+cjCqaMZE2hPcYWpleHzcNnhpZw4IY9rX9rKH1/aypI1DVw1v5yJSQzjumJxXtzUzKOrG3lzSwsWN/S77PBSjh6Tw/gxoxXEyJAqDPq45uQx2OxCTEu9PkAPgMcYRudlMDovQyFqmsjP9DF9EGZ3jc7L4MNzSrlkdgkrqt0eYM9scPufFWZ6WTQ+j5Mm5Cuw7UXBk8gBsLEY8fb0++PctjS7s3OW3EtlU+O+H+DxQGbQDX0yMt2vzKB7xrHt1xPbc7Kzaa6tgc4O6Gh3l2R1tLu3m6LYzsT1jg73MrbzmYd2nTPVAG74c/I5bq+mKTMwvsH7X10zbhJm3GewF30M+/LT2Kcfwd76N+wd12HmHeMGVFNm9vwj0V1TRfzn34G1KzHnfBBz7geHbT+nwWAKiuHwYszhRwOJWXHVW7BrV1E8djx15WP1D26KWGtZsqaRf75WRXNnHN5wZxGW5/iZURbsWa4zKjdwQN+jDY0d/NeSjcTilh8vHpvUD+UydDJ9HuaUu6enrqgo0gemhDH5Gfz4lLEsXRvln69V8dWH1nG2U8iHZpcc0CyGdfXtPLq6kSfXNtLUGacky0doVjGLJ+YfEssaJb15PYaKwiwq2xpSXYpIWvMYw4yyLGaUZXHFkWW8srmFJ9Y28sDKeu5ZUc/Y/AAnTMjnhPF5lGYfXLNWB0rBk8gBsDf+kcqXnoL5x2FOOhszbnJq62moxT56D/bJh9wZSofNo/D4U2no6MDsFCplQWbiekYQfL5+f/AsqKigdQAfSGx3txtAdbbvCKO2B1VdnYyYu4Cq2L7HSTaTGcQsOg0WnYbduNadBfXCUuyLT0L5KPeMeCPHsO3630NbK55Pfxszd+HQFzrMGGPc/k9lFQQrKjD68JoSW5o6+eOLW3lrWyszSoN8av4I8otKeGrZBt6tauWVzS08viYKQH6Gl+mJIGpGWZAJhZn4+vk/3Gvq2vn+4xvxGvjvU8cyNn/3Pl8iw50xhpMm5jN/VA7Xv1HNfSvqeXZ9E588soxjxuT2+9/P5s4YT6+L8tjqRt6ra8fnMRw1OodTJxcwe0SWZpaIiAxjfq+HhWNzWTg2l6aOGM+sd2dA3fBGNTe+Uc3MEVmcNCGPY8bmHpLLLxU8iewnu3Uz9tkl+CdOoeuVZ93m1RMdN4CadyzGP3Sptt262T0L2fOPQyzunsXtjAsxYyaQU1FBNIUf/o3PBz4f7GFpmm9EBaQ4nDBjJmA+dBX2wo9hX33G7QV12z+wgK9iDHz5h5iKsSmtUaQ/uuOWu5bXcevbNfg8hk8vGMFpkwvwGJM4I04R508vwlrL5mgny6rbWFbVyvLqNl7Y6J4tJ9PnnvVsRmmQGWVZTC0O9tkwM1LTxg+f2EjQ5+FHi8dSkadZGnJwy8nw8pmjylk8KZ8/vbSVnz1dydyR2Vw5fwQj99BsOW4t71a5jcKf29hEZ8wyviCDT84r44QJ+cPyjFIiIrJ3uRlezpxayJlTC9na1MnSdVGWrm3kdy9s5c8vb2PB6BxOHJ/PERWHTusOBU8i+8k+EAa/j9If/pattXXY5x/HPn4/9u//hw3/3e0bdPwZmKKSwath3SriD90Brz0PXh/muFMxp12AKS0ftGMezExGBuaYxXDMYuzm9dj3ljPi3A+wNdqU6tJE9mlVbRt/eHEra+s7WDgmhyuOHEHxHpoRG2MYnZ/B6PwdZ+epbe1yT9meCKNufbsWSy0eAxMLM3danrd1Yz3/tWQjBZlefrR47E7NgUUOdk5JkF+eMZ4HVtZz05s1fOH+tVw0s5j3zyjq2aemtYvHE43CtzZ3ke33sHhiPosn5TO5KFNLkEVEDhHluQE+OKuEiw8rZmVtO0vXNvL0+iaeWd9EXoaXy4/q5pQxB//fUQqeRPaDrarEvvgkZvG5eAuLMW0dmMXnYk86G5a/SfyJ+7EP3IZ98HY4/Gg8J58NUw9Lyh+a1lpY8RbxB2+H5W9CMMud3XTKuZi8PZyzWAbMjBqHGTUOT04uKHiSNNbWFefmt6q5L1JPQaaPbx8/qufMZQNRnOXnuHF+jhuXB7hnPYvUtLGsqo1l1a08uLKBe1bUA+A1hpG5fq5ZPGaP4ZbIwczrMZw7rYhjxuby91eruPmtGpaujfLhBfD48s28vqWFuIVZI7K4ZHYJC8fkkuFTf0ARkUOVMTvOuPeJeSN4rbKZpWujrKltgTEFqS5v0Cl4EtkP9v7b3BlGp79/p+3G44GZR+CdeQS2eiv2yYewzzxK/LXnoGKsuwzv6BMxmQM/45ONx+D1F93Aaf17kF+IufByd1bVIXiGNZH9Fe2IURntJJCX3DOcpMJrlc386aWtVLV0c+aUAi47vJTsQHKW7mQHvMytyGFuhXvWs65YnPfq2lle1UanN5MzxmdQkKk/I+TQVpzl5xuLRvFaZTN/fnkbP1+ykuKgj4tmFnPyxPw9LsETEZFDl89jWDA6lwWjcw+ZMyDqL0aRAbJVW7AvPOGehS1/zzOMTGk55qKPYs+7xD172uP3YW/6E/bO6zDHLMaceBamfNS+j9fVhX3hCezD/4Ftm6FsJOayz2IWnoTx6w9akb7EraW6pYtNjZ1sinayKdrBpsZONkc7aexwu9l7PRtYOCaHs6YUMqMsOKyWvjS2d/O3V6t4al2U0XkBfnrqWKaXZQ3qMf1eD9NLs5hemnXI/JEk0l9zK3L43TlZdATyyeqKqlG4iIhILwqeRAbIPtD3bKc9MYEMzLGnYI9ZDGsibh+opQ9il9wLM45wl+HNmofx7DxLwba3uk2uH70bGupg7EQ8n/oGzF24274ih6qO7jiVTZ2JgKmDTVE3XNoc7aQzZnv2y8vwMjovwILROYzODzAyJ8C6Fg93v72ZZ9Y3Ma4ggzOnFHDihPw+G2mnC2stT6yN8o9Xt9HWHeeSWSVcOLMIvzd9axY5VAS8HsaPyKWyUsuzRUREelPwJDIAtnor9vnH3SVzBUX7fkAvxhiYNA0zaRr24o9jn3oE++RDxH//Yyguw5x0FubYU4g1BonfdSP2ifuhtQWmzcbzsS/C9MOH1YwMSY64tdS2drOlqZMtTV1saeqksqmTrU1ddLOOOSMyWTgml5llB/epuKMdMbZtauCN1Q1sinawOerOZKpq7mJ7vGSAETl+RucFmFOezai8AKMTX3l9LAm7oKKC8ycHeWpdlAdW1nPty9u4/o1qTpqYz5lTChiTnzGkz3FftjR18qeXtvLm1lamlQT57NHljE2zGkVEREREdqXgSWQA7AO3gceLOaN/s532xOQVYs65GHvGhfDmi8Qfvx97+7+wd9/MFo/BdnbCEUfjOeNCzISpSape0lUsbqlp7eoJlrY0dbKl2b2+tamLrviOmTt+j6E818/I3ACZmZk8trqWB1Y2kJfhZcHoHI4Zk8vs8mz83uEZQnXHLZXRTtbWt7OuoYP1DR2sre+grq27Z5+A1zA6L4BTHOTkifmMyQswKi9ARV6AwABn/mT6PJw2uYBTJ+UTqWnngZX1PLyqgfsj9cwekcVZUwtZMDonpaFeLG65e0Udt7xVg9cYrpo/gtOnFOBREC0iIiIiw4CCJ5F+sjXb3NlOJ5yJKShOypjG54N5x+Kddyx283rskw+SFQzSdvRizMjRSTmGpIeO7jgb61t5s7LZDZiaO9na1EllUxfbmjvpju/YN+A1jMwJUJEb4MiKHEbmBhiZCJuKs3w9gUNFRQVrNmzitcpmnt/YzLPrm3hsdSNZfg/zR+WwcGwuc0dmp+2ZlBrau1lXvz1ccoOmjY2ddCeCNp8HxuRnMKc8i/GFGRw+sYKsrmZKsn1JD12MMUwrDTKtNMjH53Xz2HuNPLSqnp8+vZniLB9nTC7g1MkFFAaH9p/N1XXt/P6FLayp7+Co0Tl8av4InUVORERERIYVBU8i/WQfuA2MwZxx4aCMb0aNw3zoKorUtHdYiMUtTZ0xGttjNLZ307DLZWNH4rI9RkN7N+3ddqfHZ3gNI3MDjM3P4KjRO8KlitwAhcH+ByuZPg/HjM3jmLF5dMXivLm1lec3NvHipmaeXBclw2uYW5HDwjE5zB+dQ5Z/6PuDdcUsm6PuzKV1DYmv+nYa2mM9+xQGfUwoyOCIkdmMK8hgQmEmo/IC+HrNNKqoKKaycvDPRFeQ6eOiw4q5YEYRr2xu5oFVDdz0Vg23vlPDwjG5nDW1kOmlg9uMvL07zq+fWMUtr24kP8PLtxaNYuHY3EE7noiIiIjIYFHwJNIPtrYK+9wSzPGnYwqTM9tJ0lt9WzevVjbTvb6TjdX1O4dK7TGiHTFsH4/zGMjP8JKf6SM/08uIkgAFme7t8eXFBLtbGZkXoDDTm/Tgwu/1cOSoHI4clcNn4pZ3qlp5fkMTL2xq5vmNTfg8hjnlWRwzNpcFo3L67Hu0P+LWEu2IUdfaTV3bjq+G1+tZVtnApsYOtvf59nkMY/MDzK3IYXxBBhMKMxhXkEF+kmpJJq/HcNSYXI4ak8vmaCcPrqrn8dWNPL2+ifEFGZw1tZATJuSROcAZZbG4+3rVt3XT0N5NfVs39e3u7e3bNkU7aWyPcfrkAj5yRCk5AZ1QQERERESGp/T7S18kDdkHbk/Mdroo1aXIIGrvjvPCxiaeXBvlja0tbG+tlOX3kJ/pJT/DR0VegOmlbqhUkAiXdlz3kRPw7HG20lCegt7rMcwpz2ZOeTZXzrdEatp4fkMTz29s4ncvtOAxcFhZFkePyeXoMTl9Lt+y1tLSFXeDpO2hUms3dW1dOwKm1m7q27t3Wiq4XVlOBmPyfBxZkc34wkzGF2YwKjcwLJugj8oL8Ml5I7h0TmlPM/I/vrSVf71exckT8zlzagF5xd1sjnbS0Oa+JtuDpPr2WM+2hjZ3Nly8j9Qyy++hINNHYdDL7BFZXLpwMuXetqF/siIiIiIiSaTgSWQfbG019tnHMItOxRSVpLocSbJY3PL2tlaWrm3k+Y3NtHfHKc3y8f4ZxSwal8s8Zxy1VdtSXeYB8RjD9NIsppdm8bG5Zayp7+C5RAj1l1e28ZdXtjGtJMicsc1U1jbuFCp1xHZPSLL9HoqyfBQFfRw2IouioK/ndlHQT3GWj4JMH+PGjDrolo32bka+oqaNB1Y28NCqeu6L1ANrd9vf5yERJvkoyfIzpTiz53Zhpo+CoJeioPt67dqLq6KikMpKBU8iIiIiMrwpeBLZB/vQ7QB9znba2tTJe+/VMD7T7tSLRtLf2vp2lq6N8uS6KPVt3WT7PSwal8uJE/KZURbsmbWU4Tu4ljgZY5hUlMmkokwuO7yUDY0dvLChiec2NnHP21soyPRQFPQxuSiTolE+irP8FAZ9FCfCpcKgb8BLyw5GpleY94l5ZTy9Lkp2bh6ezpZeodLeZ8CJiIiIiBwKFDyJ7IWtq8Y+8yjm2FMwxaU73Vfd0sW3H91AXVs3RUEfp08u4LQpBRQN8VmvpP9qW7t4cl2UpWujrG/owGtg3qgcThyfx/zROQS8h16gMjY/g7GzMgjNKhnSpYAHk4JMH+dOK9LrJyIiIiLSB31CFtkL++AdYMGctfNsp5bOGD96YhPt3XG+c5rDQ29v4pa3awi/U8PRibNezSwb3LNeSf+0dsV4fkMTS9dFeXtrKxZwSjL51PwRHDc2N2kNtkVERERERGR3+sQlsge2rgb7zCOYY07GFJf1bO+KWX769GY2RTv4/sljOHPOKI4qNWxp6uShVQ08trqBZzc0MS4/gzOnFnDChLyUnML+UBaLW17f0sKTa6O8sKmJzpilPMfPxbOKOWF8PhV5gVSXKCIiIiIickhQ8CSyB/ahO8BazFkf2LHNWn7/4hbe2trKFxeOZE55ds99I3MDfGxuGR+aXcLT692zXl378jaue72akybmcebUQsbmZ6TiqRwSYnHL6rp2blmxkgff3UJje4zcgIfFE/M5YUIe00o0A01ERERERGSoKXgS6YNtqMU+/QjmmMWYkhE9229+q4ala6N8aHYJJ0/M7/OxGT4Pp0wqYPHEfFbWtvPAynoeea+RB1Y2cNiILM6aWsBRo3MHvRl5e3ecymgnGfmdg3qcVGnpjLGytp3l1a2sqG4jUtNOe3ccv9dwZEUOJ03IY25FDn6vwiYREREREZFUUfAk0gf70J1g45gzd/R2euS9BsLv1HLKpHxChxXvcwxjDE5JEKckyMfndvPo6kYeXlXPz56udJuRTyngtMkH1ozcWktDe4xN0Q42NXayKep+bW7soLq1GwCvZz1Hj87h7KmFzBimfaestWxr7mJFTRvLq9tYUd3G+oYOLOAxMK4gg5Mm5DGtNMg58ybTXFed6pJFREREREQEBU8iu7ENddinHsYcfRKmtByA1yqb+dNLWzliZDafXlA+4PAmP9PHRTOLuWB6Ea9WNvPAygZueauG8NtuM/J9hUKxuGVrcxebGjt6wqVNjR1sjnbS0hXv2S/TZxiVl8GMsixG5weoyA2wud3L3W9tHlZ9p7picdbUd/TMZlpR3UZ9ewyAoM+DUxpk4ZhcppUGmVqSudNzycv005yqwkVERERERGQnCp5EdmEfvhNi3ZizQwCsqWvnf5+uZFxBBt9YVHFAS+S8HsOC0bksGJ1LZbSTB1fVs2RN406h0NEmhzfWNCYCJncm09bmTrp35EsUBn2Mzgtw/Pg8RucHGJ2Xwai8ACVZvt3Cq4qKCt43KchT63buO3Vyou/UmDToO9XY3s2KmraekGlVbTtdcQtAeY6fOeXZTCsNMr00yJj8DLyDvExRREREREREkkPBk0gvtrEe++RDPbOdqlu6uGbpJnICHq4+cXRSZwlV5AX4xLwRXDqndKdQ6NqXtwHuErKRuQFG5wU4anQOo/PdcGl0XoDswMDqyPB5OHVyAadMSvSditTz8HuN3L+ygVm9+k4NRaBjraWyqYvl1a28W9XGe/Ub2FDfCoDPA5OKMjnbKWRaSRCnNHhASxFFREREREQktfSJTqSXHbOdPkBzZ4wfPrGRzu44Pz1tHMVZ/kE55q6hkAnmkdXdzIicQNIbY/fuO/Wxed08trqRh1bW879PV1Ic9HFaEvpO7SoWt6ypb2dZVRvLq1tZVt1GY2LZXG6GlyPGFHLSOHdG0+TiTAJeT9KOLSIiIiIiIqml4EkkwUbrsU8+iDnqBLqLy/mfJzaxpamT7580hrEFg78cbXsoVFFRSmVl16Afr6BX36lXduk7tXBsLmdN2b9m5G1dcVbWtrGsyg2ZVta00d69Y9nc3JHZzCjLYkZpkFF5AUaNGkVlZeVgPEURERERERFJMQVPIgn24f9AVzecFeJ3L2zlnW2tfPmYkcwuz051aYPK6zEcNTqXo3bpO/XM+ibGFWRw5pQCTpyQT9Df90ykhvZulle1say6leXVbayuayduwQDjCzNYPKmAGYn+TIM1a0xERERERETSk4InEcBGG7BLH8AcdQI3bQ3w5LpaLp1TwokT8lNd2pDa3nfqw7v0nerdjDwWbGXp6gaWVbexrKqNyqZOAAJew9TiTC6cUcyMMnc530B7UYmIiIiIiMjBRcGTCGAfcWc7PXL4+7j93VpOn1zARTOLU11WymT6PJw2uYBTJ+UTqWnngZU7mpHDWgByAh6ml2Zx6uR8ZpRmMakoA7/6M4mIiIiIiEgvCp7kkGebGrFPPMBrR7+fP0c6mFeRzafmjxhwb6ODkTGGaaVBppUG+fi8bp5aF6WsqJCKQCej8wN49BqJiIiIiIjIXih4kkOefeQuVmeU8ousBUzIz+Drx43C61GgsquCTB/nTSuioqJCzcBFRERERESkX7QuRg5ptinKtuee5b/nXklepo+rTxyzxybaIiIiIiIiIjIwmvEkh7SmR+7lR86ldPkz+fFJYygM6kdCREREREREJFn0KVvSmn3zJar/+iTxMRMxRx6HKRmRtLE7Gxv5n6pituUWc82JYxmTn5G0sUVEREREREREwZOksfiTD2FvupbO3FzsS09j77gOJkzFzF+EmXcspqhk/8e2lt88+A7L8sbz1ekBZo7ISmLlIiIiIiIiIgIKniQNWWuxd9+EvT8Ms45k5A9+xZbIcuwrz7hf4b9jw3+HydMxRy7CzDsGU1A0oGPc8PJmnokV85GOdzl+7oWD9ExEREREREREDm0KniSt2O5u7A1/wD63BHPcqZhLP4MnM4gpLceceRGceRF26+YdIdS//4K99a8wZSZm/nGYucdg8gr2eozbX9/EnauaOX3zc1zwoZOH5omJiIiIiIiIHIIUPEnasO1txP/8v/DOa5hzP4g59xKMMbvtZ8pHYc65GM65GFu5wQ2gXn4Ge9O12Jv/AtNmuf2g5i7E5OT1PK47bnlwZT3/eK2KI+sjfDKvGs/o8UP4DEVEREREREQOLQqeJC3YaAPx314DG9ZgLvssnuNP79fjTMVYzHkfwp57CWxe7wZQrzztzpq66U8wfQ4cuYiXRszh+uVNVDZ1Mi/Qwlfevh7/1b8c5GclIiIiIiIicmhT8CQpZ6sqif/6B9BYh+ez38HMWTDgMYwxMHo8ZvR47Ps+DBvXYF9+hhXvrub6N7tYXlDL6O5GvjOykyMf+BMcfiRGs51EREREREREBpWCJ0kpu3Yl8d/9CGwcz1d+jJk07YDHNMawpXAMN5SdwrPtR1Hgh0/HV3PyW3fifaYaAM85Hzzg44iIiIiIiIjI3il4kpSxb79C/Nr/hbwCPF/8AaZ81AGPGe2IEX67hgdX1eM1hg/OKuZ904sJ+qdhQ2fC2pUUBTOor5iQhGcgIiIiIiIiInuj4ElSIv7Mo9gb/gCjJ+D5wn9h8gsPaLzOWJz7VtRz+7u1tHXHOWVSPpfMLqUouOMtbjwemDSNYEUF9ZWVB/oURERERERERGQfFDzJkLLWYu+/FXv3zTDjCDyf/iYmM2u/x4tby1Protz4RjXVrd0cWZHN5UeUMbYgI4lVi4iIiIiIiMj+UPAkQ8bGYtib/4x96iHM0SdhLv8cxuff7/He2trCv16vYnVdB5OKMvjCwpHMLs9OYsUiIiIiIiIiciAUPMmQsB0dxP/6c3jzJcyZF2EuuMw9E91+2NDQwb9er+LVyhZKs3x8+ZiRHD8+D89+jiciIiIiIiIig0PBkww62xwl/vsfw5oI5pIr8Zx8zn6NU9fWzc1vVrNkTSNBn4fLjyjlHKeQgNeT5IpFREREREREJBkUPMmgsjXbiP/mB1BTheeqb2LmHjPgMVo7u7n5rWruWlZHzFrOdgoJHVZCXoY3+QWLiIiIiIiISNIoeJJBYzesJv7ba6CrC89XfoSZMmNAj2/qiPHoew3cu3INda2dHDcul8vmlFKeGxikikVEREREREQkmRQ8yaCwy94g/qf/gaxsN3SqGNvvx25o6OC+SD1PrG2kM2aZP7aQb00fiVMSHMSKRURERERERCTZFDxJ0rU88SDx3/4QRo7B84XvYwqL9/mYuLW8srmZeyP1vLW1lYDXcOKEPM5xilg4YwKVlZVDULmIiIiIiIiIJJOCJ0kqu341db+4GpxZeD7zHUxW9l73b+2KsWR1I/dF6tna3EVxlo/LDi/ltMkF6uEkIiIiIiIiMswpeJKksls2AOC59NN7DZ0qo53ct7KeJasbae+OM700yEcOL+XoMbl4PWaoyhURERERERGRQaTgSZKrOepe5ubvdpe1lje2tnLfijperWzB64HjxuVxrlPE5OLMIS5URERERERERAabgidJrqYoeLwQ3DHbqb07zhNr3OV0m6KdFGR6+eCsEk6fUkBhUG9BERERERERkYOVPvVLcjU14snLx3g8VDV3cf/Keh5d3UBLZ5xJRZl8aeFIjhuXi9/rSXWlIiIiIiIiIjLIFDxJUtnmKJGyadz91CZe3NQMwMIxuZzrFDKtNIgx6t8kIiIiIiIicqhQ8CRJ9V67n29VnEvutlYumF7EmVMLKc32p7osEREREREREUkBBU+SVBtjGQD89LRxjM7PSHE1IiIiIiIiIpJKarQjSRWNuUvpCtQ0XEREREREROSQp+BJksbGY0StDx+WbL/eWiIiIiIiIiKHOqUDkjwtzUT92eR742oiLiIiIiIiIiIKniSJmhqJ+rMpCOhtJSIiIiIiIiJp2Fw8FAoVAzcAk4BOYBXwqXA4XB0KhY4G/gwEgXXApeFwuCpVtcoumqLujKdMb6orEREREREREZE0kI5TUyzws3A47ITD4VnAauCnoVDIA9wIfDYcDk8FngJ+msI6ZVfNUaL+HAqzdTY7EREREREREUnDGU/hcLgOWNpr0wvAp4F5QHs4HH4msf1a3FlPHx/K+mTPbFMjjYFyCnOzUl2KiIiIiIiIiKQBY61NdQ17lJjl9AhwD7AZ+Hg4HD671/2twOhEWNUf6ftkDwK1t/yNMzZN5Mqjx3LFosmpLkdEREREREREhsYezzCWdjOedvE7oBn4PXBBMgasrKxMxjApV1FRkdTnkozxaiqrgIkU5mSmXW0aT+Olw3jpXJvG03ipGkvjabxUjpfOtWk8jZfK8dK5No2n8VI11nAYL5UqKir2eF869ngCIBQK/QKYAlwcDofjwAZgXK/7S4D4AGY7ySCLtnQAUJAVSHElIiIiIiIiIpIO0jJ4CoVCP8Ht6fS+cDjckdj8KhAMhULHJW5fBdyWivqkb43t3QAUBv0prkRERERERERE0kHaLbULhUIzgW8DK4HnQqEQwNpwOHxBKBS6DPhzKBTKxG0sfmnKCpXdRDvjABRk+aGzLcXViIiIiIiIiEiqpV3wFA6H32UPTanC4fBzwKyhrUj6K9rlXhYGA7R3prYWEREREREREUm9tFxqJ8OPtZZo3IvBkhdMuzxTRERERERERFJAwZMkR0cbjd5Msk0Mn0dvKxERERERERFR8CTJ0hQl6s8h32tTXYmIiIiIiIiIpAkFT5IczVGa/NnkBfpszyUiIiIiIiIihyAFT5IczVEaA9nkZ6i/k4iIiIiIiIi4FDxJUtimRqL+bHKz/KkuRURERERERETShIInSYp4NEqTP4v8nGCqSxERERERERGRNKHgSZKiubmVuPGSl52R6lJEREREREREJE0oeJKkaGpuAyA/Uz2eRERERERERMSl4EmSorGtC1DwJCIiIiIiIiI7KHiSpGjsjAOQl+FNcSUiIiIiIiIiki4UPElSRDstoOBJRERERERERHZQ8CRJEY25b6X8TAVPIiIiIiIiIuJS8CQHzHZ3E8VPJjECXr2lRERERERERMQ1oE7QoVDoCOBq4HigAFgQDodfC4VCPwGeCofDDyW/REl7zVGigWzyvfFUVyIiIiIiIiIiaaTf01NCodBxwPPANODmXR4bB65KbmkybDRHifqzyfOnuhARERERERERSScDWRf1U+BhYCbwlV3uew2Ym6yiZJhpaqTRn6PG4iIiIiIiIiKyk4EET3OBP4XDYQvYXe6rAUqTVpUMK7a5yZ3xlKkpTyIiIiIiIiKyw0CCp3Ygaw/3jQQaD7wcGY5sUyNN/mzyczJSXYqIiIiIiIiIpJGBBE/PAF8KhUK911Ntn/n0CeDxpFUlw0p7UxOdXj95OcFUlyIiIiIiIiIiaWQgZ7W7GngWeBO4HTd0ujwUCv0fMA+Yn/zyZDiINrcDkB/UUjsRERERERER2aHfM57C4fCbwCJgG/BdwACfS9x9QjgcjiS/PBkOGls7AdRcXERERERERER20q8ZT6FQyA+cBbwVDocXh0KhTKAIaAiHw62DWaCkv2h7F2RBfuZAJtCJiIiIiIiIyMGuX0lBOBzuCoVCYeAMYG04HG4HKge1Mhk2op1uqy/NeBIRERERERGR3gbSXHwNUDZYhcjw1RgzAORnKngSERERERERkR0GEjz9DPhuKBQqHaxiZPix1hKN+/ARJ+gbyNtJRERERERERA52A2nKczJuX6e1oVDoBWAL7pnttrPhcPjyZBYnw0BbC1FfFnmeGMaYVFcjIiIiIiIiImlkIMHTcUAXUA1MSnz1Znd7hBz8mqJE/dnkqa+4iIiIiIiIiOyi33FBOByeMJiFyDDV7AZP+QEtsxMRERERERGRnSktkAPT1Eg0kE1epqY8iYiIiIiIiMjOBpQWhEKhLODjwAm4/Z7qgCeAf4bD4bbklyfpzjZHifpLyMsKpLoUEREREREREUkz/Z7xFAqFyoHXgN8CRwJZicvfA6+FQqERg1KhpLWuaBOtviB5OcFUlyIiIiIiIiIiaWYgM55+BhQCi8Lh8LPbN4ZCoWOAO4D/BT6a1Ook7TU1twCQn52R4kpEREREREREJN0MpMfTmcC3e4dOAOFw+Dnge8DZySxMhofGlk4A8jO9Ka5ERERERERERNLNQIKnHKByD/dtStwvh5jG9i4A8jPUXFxEREREREREdjaQ4CkCXLaH+y4FVhx4OTLcRDviAORqxpOIiIiIiIiI7GIg01R+AVyfaCJ+M7AFKAc+CJzCnkMpOYhF3QlP5GcoeBIRERERERGRnfV7xlM4HL4RuAo4DPgbcD/wd2A2cFU4HL55UCqUtBaNezBYcgIKnkRERERERERkZwNZakc4HP4LUAHMBBYlLkeFw+G/DkJtkuZsVydRk0GuieH1mFSXIyIiIiIiIiJpZsAdocPhcBxYPgi1yHDTFKUxkE2uN57qSkREREREREQkDfV7xlMoFPpVKBS6YQ/33RAKhX6evLJkWGhuJOrPId+v2U4iIiIiIiIisruBLLU7D3hkD/c9DLzvgKuR4aU5SpM/i7zMAU+cExEREREREZFDwECCp1HAhj3ctylxvxxCbFOURn8O+UF/qksRERERERERkTQ0kOCpHpi8h/smA80HXo4MJ/GmKM3+LPJyMlNdioiIiIiIiIikoYEET48B3wuFQiN6b0zc/g7waDILk/TX3NRC3HjIywmmuhQRERERERERSUMDac5zNfAysCoUCt3HjuV15wDtwPeSX56ks8bmdvBAvno8iYiIiIiIiEgf+j3jKRwOrwPmA3cBJwFfSlz+B1gQDofXJr88SWfRti5AwZOIiIiIiIiI9G1AiUEifPrI4JQiw020oxuyIS/Dm+pSRERERERERCQN7fdUlVAolA9MAbaGw+FNyStJhovGTgtAfqaCJxERERERERHZ3V6X2oVCodNDodBP+9j+XaAKeBFYHwqFbg6FQlpvdYiJxty3j2Y8iYiIiIiIiEhf9hUWXQXY3htCodCpwI+At4G/AdOBTwGvAr8chBolDdl4nKj1EySG3zuQkyOKiIiIiIiIyKFiX8HTEbghU28fwz2L3enhcHgrQCgUAvgQCp4OHa3NRP1Z5Hljqa5ERERERERERNLUvqaqlAGrd9l2KvDM9tAp4X5gajILkzTXFCXqzyZfCyxFREREREREZA/2FTw1Adnbb4RCoSlAMfDCLvtFATX6OZQ0NRL1Z6u/k4iIiIiIiIjs0b6CpxXA+b1un4/b8+mRXfabAGxLYl2S7pqjNAayyQtqypOIiIiIiIiI9G1fqcGvgDtDoVARbrD0Udym4s/ust9ZwJtJr07SVrypkah/NHlZmakuRURERERERETS1F5nPIXD4buALwHzgY/gLrH7QDgc7jnTXSgUKgdOAR4YtCol7bQ3NdPt8ZGfG0x1KSIiIiIiIiKSpva5TiocDv8W+O1e7t8KlCSzKEl/jc1tAORnBVJciYiIiIiIiIikq331eBLpU7SlE4D8TPV4EhEREREREZG+KXiS/dLY0Q1Ars5qJyIiIiIiIiJ7oOBJ9ku0Iw5AvoInEREREREREdkDBU+yX6IxA0BepoInEREREREREembgifZL9G4Fz9xgj69hURERERERESkb0oNZMBsRzuNnkzyPDGMMakuR0RERERERETSlIInGbjmKNFADnlem+pKRERERERERCSNKXiSgWtqJOrPIi+g2U4iIiIiIiIismcKnmTgmqJE/TnkZ/pTXYmIiIiIiIiIpDEFTzJgtjlK1J9NXlYg1aWIiIiIiIiISBpT8CQD1hWN0ubLJD83M9WliIiIiIiIiEgaU/AkA9bY3ApAXnYwxZWIiIiIiIiISDpT8CQD1tTSAUB+pi/FlYiIiIiIiIhIOlPwJAPW0NYFQF6mN8WViIiIiIiIiEg6U/AkAxbtiAOQn6HgSURERERERET2TMGTDFjUnfBEnoInEREREREREdkLBU8yYNG4Bw+WHAVPIiIiIiIiIrIXCp5kQGwsRhQ/OaYbjzGpLkdERERERERE0lhanpYsFAr9ArgQGA/MCofD7yS2TwWuA4qBWuAj4XB4VarqPCS1RIn6s8n32lRXIiIiIiIiIiJpLl1nPN0FHA+s32X7tcAfwuHwVOAPwJ+HuC5paiLqzybPn+pCRERERERERCTdpWXwFA6HnwmHwxt7bwuFQmXAXOCWxKZbgLmhUKh0qOs7pDU30ujPVmNxEREREREREdknY236LpkKhULrgHPC4fA7oVBoHnB9OBye2ev+ZcCl4XD4tX4Omb5PdphofeYxzn2ylZOnlfPdCxekuhwRERERERERSb09NoFOyx5Pg6mysjLVJSRFRUVFUp9Lf8frWr+OJv94Ap74HvdPVW0aT+Ol+3jpXJvG03ipGkvjabxUjpfOtWk8jZfK8dK5No2n8VI11nAYL5UqKir2eF9aLrXbg43AqFAo5AVIXFYktssQaY62YI2HvNysVJciIiIiIiIiImlu2ARP4XC4CngDuCSx6RLg9XA4XJ2yog5B0ZYOAPKzAimuRERERERERETSXVoGT6FQ6LehUGgTMBp4LBQKvZu46yrg86FQaCXw+cRtGUKNrZ0Aai4uIiIiIiIiIvuUlj2ewuHwF4Av9LF9BXDU0Fck20U7uiEI+ZkKnkRERERERERk79JyxpOkr2ine2JAzXgSERERERERkX1R8CQD0hhzz5CYl5GWk+VERERERETk/9u79yBL87I+4N++90zPnN5lEXBhgUUXFUpFkPISvF9IokRzqVfxAmqM18SIRWlMKsGYkrLUClYCRBKiEEXMz4qBxBsWFYmXSlSyksSQyMVdFlnktrv9ds9Mz0x3n/xxzs72zJzre7r37dN8PlVbvX0uzzzvec7zO+95+j3nhRPE4ImJdbvd1N3lnM1eVpYW2k4HAAAAOOEMnpjc7qXUS2fTWTxoOxMAAABgDhg8MbntrdQrZ9NZaTsRAAAAYB4YPDG5nTr16rlsrnraAAAAAOOZIDC57Tr1ykY6Z3yxOAAAADCewRMTO+gPnjbPrrWdCgAAADAHDJ6Y2MXt7ewtLqdz/kzbqQAAAABzwOCJidU7F5MknY31ljMBAAAA5oHBExOrL15Jkmyu+44nAAAAYDyDJya2dWkvSbK5vtRyJgAAAMA8MHhiYvXlgyRJZ83gCQAAABjP4ImJ1b0DntJZ81E7AAAAYDyDJyZWHyxlNftZX15oOxUAAABgDhg8MZHu3tXUi2vpLOxnYcHgCQAAABjP4InJ7NTZWtlIZ7nbdiYAAADAnDB4YjLbdeqVjWyuOtoJAAAAmIzBE5PZqbO9suGLxQEAAICJGTwxke72VrZWzqVzdrXtVAAAAIA5YfDERK5sb2d3eS2dc2faTgUAAACYEwZPTKSuLyZJNs8bPAEAAACTMXhiIlsXd5MknTMrLWcCAAAAzAuDJyayffFqkmRzbanlTAAAAIB5YfDERB66fJAk6awbPAEAAACTMXhiIvXVbpJkc2255UwAAACAeWHwxETqg8Usdg+yseopAwAAAEzGFIGxugcHqbvLOb+wn8WFhbbTAQAAAOaEwRPjXbqYevlsOksHbWcCAAAAzBGDJ8bb3kq9ci6bK20nAgAAAMwTgyfG29lKvbqRzpoz2gEAAACTM3hivJ06Wysb2Ty72nYmAAAAwBwxeGKsvbrOzspGOmfX2k4FAAAAmCMGT4y1vX0hSdI5f7blTAAAAIB5YvDEWPXOpSTJ5oYjngAAAIDJGTwxVn3xapJkc92XiwMAAACTM3hirK3dvSRxVjsAAABgKgZPjFVfOUiSdNaXW84EAAAAmCcGT4xV7/eeJo54AgAAAKZh8MRYWwdL2chelhcX2k4FAAAAmCMGT4zUvXI59dJ6Oov7bacCAAAAzBmDJ0bbqVOvbGTT1zsBAAAAUzJ4YrTt3uCps+qpAgAAAEzHNIHRtrdSr5xL54xDngAAAIDpGDwx0sF2ne2Vs9k8u9Z2KgAAAMCcMXhipIvb29lbXE7n/HrbqQAAAABzxuCJkba2d5MknXNnW84EAAAAmDcGT4xUX+wNnjbXfccTAAAAMB2DJ0baurSXJOmsL7WcCQAAADBvDJ4Yqb5ykCTZXHPEEwAAADAdgydGqq/2fm464gkAAACYksETI20dLGUt+1lb9lQBAAAApmOawFDdg/1sZyWdhb22UwEAAADmkMETw13YSb2ykc5St+1MAAAAgDlk8MRwO3W2VjbSWV1oOxMAAABgDhk8Mdz2VurVDWe0AwAAABoxeGK47TrbKxvpnF1tOxMAAABgDhk8MdTu9nZ2l9bSObfedioAAADAHDJ4Yqi6vpAk2Tx/tuVMAAAAgHlk8MRQ9YXLSZJNH7UDAAAAGjB4Yqiti1eTJJ31pZYzAQAAAOaRwRND1Zf3kiQdZ7UDAAAAGjB4Yqi6N3fK5pojngAAAIDpGTwx1NbeQpa6B9lY9TQBAAAApmeiwEDdbjd1dyWdhb0sLCy0nQ4AAAAwhwyeGOzyburl9XSWDtrOBAAAAJhTBk8Mtr2VemUj51faTgQAAACYVwZPDLaznXrlXDZXfbE4AAAA0IzBE4Pt9I542jzjkCcAAACgGYMnBrpab2Vn5Ww6G2ttpwIAAADMKYMnBtrevpAk6Zw/23ImAAAAwLwyeGKgens3SdI5t95yJgAAAMC8MnhioPrSlSTJ5vpyy5kAAAAA88rgiYG2Lu0lMXgCAAAAmjN4YqD6ajdJsrm21HImAAAAwLyau8NZqqp6epLXJ7ktyceSvKiU8u52szp96qu9n+cNngAAAICG5vGIp59J8qpSytOTvCrJa1rO51Squ0s5l6tZWlxoOxUAAABgTs3V4KmqqscleXaSN/YvemOSZ1dV9QntZXX6dPf2srWwls7iftupAAAAAHNsrgZPSe5I8oFSyn6S9H/e37+co3JhO/XKRjpz90FMAAAA4CRZ6Ha7becwsaqqnpPk35VSnnnosncm+aZSyt0ThJifjW3RlXvfkxf+3B/kyXc8Ia/49i9rOx0AAADgZBv6PT3zdkzL+5M8saqqpVLKflVVS0lu718+kfvvv//Ykns03X777Ue6LYfjdf/sPalXNnJmqdvo3zjO3MQTb57jneTcxBOvrVjiiddmvJOcm3jitRnvJOcmnnhtxZqHeG26/fbbh143Vx+1K6V8OMk7krywf9ELk/xxKeUjrSV1Ch3UdbZXNtI5u9p2KgAAAMAcm7cjnpLku5K8vqqqf5LkwSQvajmfU+fC9k72Fx+bzXPrbacCAAAAzLG5GzyVUv5fks9pO4/TbGv7YpKk09loORMAAABgns3VR+14dNQXdpMkm2dWWs4EAAAAmGcGT9yk3t1Lkmyuz90BcQAAAMAJYvDETbYuHyRJOmtLLWcCAAAAzDODJ25SX+0mMXgCAAAAZmPwxE3qg6Wsd/eytuzpAQAAADRnssB1ut1u6u5KOot7bacCAAAAzDmDJ6536WLq5TPpLHXbzgQAAACYcwZPXG9nK/XquXRWFtrOBAAAAJhzBk9cb7vO1spGNtd9sTgAAAAwG4MnrrdTp17ZSOfMatuZAAAAAHPO4Inr7G7VubK0ms1z622nAgAAAMw5gyeus7V9MUly/vzZljMBAAAA5p3BE9epd3aTxBFPAAAAwMwMnrjO1qUrSZLNteWWMwEAAADmncET19na3U8SZ7UDAAAAZmbwxHW2r3aTJJ01gycAAABgNgZPXKfeX8hy9yBnVzw1AAAAgNmYLnCdre5yzi9czcLCQtupAAAAAHPO4Ilrulevpl5cz+biQdupAAAAAKeAwROP2KlTr2yks9J2IgAAAMBpYPDEI7a3Uq9sZNMXiwMAAABHwOCJR+z0Bk/n15fbzgQAAAA4BQyeuOZqXefCytlsbqy3nQoAAABwChg8cU1dX0iSdM6faTkTAAAA4DQweOKaemc3SbJ5/mzLmQAAAACngcET19QXLydJNs84rR0AAAAwO4Mnrtm6tJck6TirHQAAAHAEDJ64pr5ykCTprBs8AQAAALMzeOKaem8hC91uzq8aPAEAAACzM3jimvpgMedyNUuLC22nAgAAAJwCBk8kSboHB6kXVtNZ3G87FQAAAOCUMHgiSXKwU6de3khnudt2KgAAAMApYfBEkuSgfihbqxvprHpKAAAAAEfDlIEkycHWQ6lXNrK5vtx2KgAAAMApYcpAkuTqQw9ke+VsNs+utJ0KAAAAcEo44okkSf1gnYOFpXTOnWk7FQAAAOCUMHgiSfLgQ3WSpNM523ImAAAAwGlh8ESS5IHtS0mSzY31ljMBAAAATguDJ5IkD+7sJkk215ZazgQAAAA4LQyeSJJsXdpLknTWDZ4AAACAo2HwRJLkwSsHSZKOI54AAACAI2LwRJLkob1kvbuX1SVPCQAAAOBomDKQJNk6WM7mwl7baQAAAACniMET6V6+nHppPZ2lg7ZTAQAAAE4RgyeSna3UKxvprC60nQkAAABwihg8kezUvcGTLxYHAAAAjtBy2wnQvm69lXrlXDbPrLSdCgAAAHCKOOKJXKq3c2VpJZ2NtbZTAQAAAE4RgyeyvX0hSbLZOdtyJgAAAMBpYvBEtnZ2kySd8wZPAAAAwNExeCL1xStJks11X/kFAAAAHB2DJ7K1u58kzmoHAAAAHCmDJ7J1tZsk2Vw3eAIAAACOjsET2d5fyHL3IGeWPR0AAACAo2PSQOqDpdyyuJeFhYW2UwEAAABOEYOnj3Pd/f1sLaxlc6nbdioAAADAKeM0Zh+nut1u0j1I6odSr2zkljVHOwEAAABHy+BpznS73Xz/L/xR9vcP8tTLH8mdlz+ap+5+OE+79OF0rl5MDvaTg4Nkf783WNrv//7w5Yf/v6/+nB/MHesrLW4VAAAAcBoZPM2h5+f+/M908n/P3J7fPXfXtctvO7iUO7t17uxu587s5M7FC3n84pUsLC4li4vJjT+XFpOFxdRbt+a2J93e4hYBAAAAp5HB05xZWFjIX/3mr82333577r///tSX93Pvg7v5swd3c88Dl3PPg7fk7vpyDvpf2XR2ZTFPvWUtT3vMeu68dS1Pu3U9d2yuZWWp99G6q/vdXPylP81jzq+3uFUAAADAaWTwNOc6a0v5jCds5DOesHHtsst7B7lv63LuefBy/uyB3dzz4OW89b0PZXevN41aXkzu2FzLnbeu5fHnVpMkt5xZbSV/AAAA4PQyeDqF1pYXc9dtZ3LXbWeuXbZ/0M1f7FzNPQ/uXhtG3X3/hTy0WydJntBZT3K5pYwBAACA08jg6ePE0uJCnthZzRM7q3neUzrXLn/w0l4+fOFqPu/Ox+QvPvjBFjMEAAAAThuDp49zt55Zzq1nlrO4sNB2KgAAAMAps9h2AgAAAACcTgZPAAAAABwLgycAAAAAjoXBEwAAAADHwuAJAAAAgGNh8AQAAADAsTB4AgAAAOBYGDwBAAAAcCwMngAAAAA4FgZPAAAAABwLgycAAAAAjoXBEwAAAADHYrntBA6rquqbkvxgkmck+f5SyisPXXc2yc8leU6SvSQvLaX8aiuJAgAAADDWSTvi6R1Jvj7JLw647qVJ6lLKJyd5QZLXVlV17lHMDQAAAIApnKjBUynlT0op70xyMODqr0vymv7t3p3k7Un+yqOYHgAAAABTOFGDpzGenOR9h36/L8kdLeUCAAAAwBgL3W73UfvHqqq6O70B0iCPL6Xs92/3uiRvv+E7nraTPK2U8pH+769O8p5Syj+fIoVHb2MBAAAAPj4sDLviUf1y8VLKs2e4+31JnpLkI/3fn5zkt2dOCgAAAIBZdDNk+HSizmo3xi8n+c4kb6+q6q4kz03ywiljDJ3AAQAAAHC0HtWP2o1TVdULk/xkkluTXElyIclXllLeWVXVRpLXJfmsJPtJfrCU8ua2cgUAAABgtBM1eAIAAADg9Jins9oBAAAAMEcMngAAAAA4FgZPAAAAABwLgycAAAAAjoXBEwAAAADHwuAJAAAAgGNh8AQAAADAsVhuOwEmV1XVbUl+PsknJbmS5N1JvrOU8pEjiP2yJD+S5NNLKX8yQ5z1JK9I8uVJdpP8t1LKd8wQ76uT/LMkC/3//mkp5VcmvO9PJfmbSZ6aQ9tVVdXTk7w+yW1JPpbkRaWUdzeJN0tNhuV36PqpajJiexvVZES8RjUZ9VhVVfW5SV6T5EySe5N8Uynlw03iJbm1H+sTk+wl+aMk31NKudQkt0O3+dkk35rkfCllZ4ZtfUySVyV5TpKrSf59KeVHZ4j3bUlekmS/v70vKaX87qh4/ZhvSnJnkoMkO0n+XinlHTP0x03xkrx/WN5N8zt0/bT9MWx7m/bHsHizrFnXbVOTvhgWL73Hf6q+GJffocsn7o1R8Zr0xph4TXvj3vSeC7v9i36olPKWpvUYFC/JPWlYj2H5Hbp+qnqM2N5G9RgRb+p6DOvPGdapm+Il+eE0X6dGrh8N1qlh29t0nRoWb+p1qqqqpyZ506GLbknSKaU8pkk9hsVLclca1GNUfoduM3E9xmzvVPUYE6vpPtXA+83QGzfFS/Jf07w3Rm5Xg94Ytr1Ne2NYvKb1+Kr+/VaSPJDkW0op98xQj5viJanTvB4D8zt0/TS9MWxbm9ZiWLyJatHk/d6Y66aKN+49xCzvR4/q/flJ4Yin+dJN8hOllE8ppXx6kvcm+fFZg1ZV9ewkn5vkfbPGSvIT6S02T+/n+I9nyGshvUb+5lLKs5J8c5LXV1U16fP2TUm+MDdv188keVUp5enp7VS/ZoZ4s9RkWH5NazIsXtOa3BRvxpoMfKz69/2FJN/br8nvZLLHcNhjfyXJD5RSPjXJZyQ5m+SlDWM9vN0v6N9mUqPivS7JH5RSnl5KeWaSf900Xv/F7qeTfHm/Hj+ayZ/PLy6lfGYp5bOS/FSSn+1f3rQ/BsWbpT+G5de0P4bFa9ofN8WbpT9u3KYZ+mJgvDTri1HxHr582t4YFe91mb43BsabsTeS5G+VUp7V/+8ts9bjxniZsR4D4iVpXo8h8V6XhvW4Md4M9RjWn03XqUHxZlmnhq4fDdepYfGarlM33a/pOlVKufdQTZ+V3j7CL/avnroeI+I1qseY/Kaux5h4U9VjWKymtRhzv6lrMSxeGtZi3HZNW4sx8abujRHxlkblPSLerek9Xl/fz+HfJPlX/aub1GNYvKb1GJXfVPUYE6tJLQbGm7I33pTp3++Num7aeOPq0iS/o35/fiI44mmOlFIeSPK2Qxf99yTfPUvMqqrW0nuyv/CG2E1inUvyoiRPKqV0k6SU8qFZYqZ3NMFm//9vSfLBUsrBJHcspfxeP6/DOT4uybOTfEX/ojcmeWVVVZ9QxvzFYFC8WWoyKF7/90Y1GbK9jWsyLL80rMmIx+o5SXYf/vfSW4jvTfJtTeKVUu7t3z+llIOqqv4wyac1zO3hN7AvS/Jl43IaF6+qqrvSe5P5NYdu+xdN4+WRvwKdT/Kh9Orx5xPmuHXo180kBzP2x03xZuyPm+IlM/XHoO2dpT8G5pcG/TFkmxr1xbB4TfpiTH6NemNYvKa9MSK/xr0xRON6DDJLPYZpWo8hsRrXY4ip6zGsP5uuU2P6/W2HbjrROjUqXpN1asT2NlqnRsRbyAz7Vv3Yq0m+McnzZ3ndGBTvKPZ1D8fr/z7Tvu4N2zvTvu6NuaV5LW66X5LHpnktBuUxSy0GbtcMtRi0vWfTvBaD4h0My3tMrE9O8qFSyrv6v/96kp+foTcGxkuyWEp526HbTVqPYfk9Nsl2pqvHsFhPSrNaDNvWx2bCWkz7fi+916KhdWn4/vFth1K6ri5N4h3l+/OTxBFPc6o/8f3uJP9pxlA/muQX+jvBs/qk9A4VfFlVVW+vquptVVU9r2mw/sJVJXlzVVXvS29i/KIZc7wjyQdKKfv9f2M/yf39y2eiJpO74bF6cg5N80spH02yWPU+5tEk3uHLz6T3BmzimgyI9aokL7th0DCxG+I9I703W6+tquruqqp+vaqqZzaN13+svjPJ3VVV3Zfk5Um+Z4pYr+3f78eSvDgz9seAeAPzniG/ZIb+GBBvpv64Md4M/TFom2bpi5GPUYO+GBavaW8MijdLb9wUb9beSPKGqqr+V1VVr66q6pbMvk7dGO+aJuvUkHizrFU3xpt1rbouXsN6DOvPpuvU2H6fcp0aFa/JOjUsXtN1auD9juh1/K+lV4O7czT7VYfjXTPDftWN8Wbdrzocb9b9qmuxmtZixP0a1WKSPKapxZh4U9diRLxGtRgWb4beeFeSJ1RV9dz+79/Y/9m0N4bFe/LDN5iyN0bFm7Yew2I17YtRj90s69Sox75JXSa6zxR1GRfvKN8LnhgGT/PrX6b3vSKvbBqgqqrPS/LZSV59RDktJXlakj8upXx2et9j8StVVXUa5rec3ncvfE0p5SlJXpCkVL2/Np1EajK5mR+rcfH6uf5Skv9SSplmp/VarKqqqiRXSim/dkS5LaV32OzrSinPTvLaTL9DfTi/TpK/m+S5pZQnJ/mBJP+x6v1Fe6xSyrf37/cPk/zklHlMG2/qmt8Yb9b+GJDfTP0xIL+p++Ooe35cvGn7Yli8pr0xIr9GvTEiv1l64wtKKZ+Z5Lnp/WV01nVqaLyG69RN8WZcqwblN8taNSi/JvUY2J9Jmu4DTNLv06xTw+J9UZr19LDt3Zwg72ny62T21/Fvy6GPPx+BYfGa7itci3dEa+zh/GbdrzqcW6N9qmH3S8PemDCPiWsxIt4XpEEtRmxvo94Ys71T16M/7P+6JK+oqurtSR6X5KE0rMeIeHuHbjZxPUbEO58p6zEiVjcNajFmW+fpPeDDTuJ7wRPD4GkOVb0vKbsrydcNOuRwCl+U3qH991S9LwN9UpK3VFX1lQ3j3ZfeQvHGJCml/EGSjyZ5esN4z0pyeynl9/vxfj/Jhcz2cYT3J3li1fscd/o/b+9f3piaTF6TAY/VfUmecuj6x+aRj2k1ifdwXd+Q5MEk3zdDbl+c5Eurqrq3X48k+T9VVT2jYbz7ktxX+l+oW3pfkviJ/W1uEu8rkzxUSvnTfryS3l+dJor3sFLKzyf5kvSOcJi5Px6OV/U++jNzfxzK70tzBP1xKN4HcgT9cSjeczJ9fwzs+fQOP2/SF0PXkIZ9MSy/H0mz3hi1vU16Y1i8v5yGvVFKeX//5+X0dvz+UmZYp4bEa7xODYn3xWm4Vo3Y3kZr1ZB4TdaqYa9fl9JsnRr5ethgnRoW7wvSbJ0aFu/iqLwbbm/j1/Gqqp6YXt+9oX/RTPtVA+I9fHmj140B8WbarxoQr/F+1YBYz0qzWgy7326a1WJkHg1qMSzeF6dZLYbFu5RmtRi1vY16o5Ty1lLK8/pDl1fmkZNQNOqNIfHe248zdW8Mife8NKjHkFjvS8O+GBLv1sz2fmPUutRkzRp7nynrMireUb8XPDEMnuZMVVUvT+/Nzdf2d+oaK6X8eCnl9lLKU0spT03vTefzSym/1TDeR5P8dvqfV61639b/uCTvaZjinyd5UlVVn9KP92lJHp/+wtswxw8neUd6n5lN/+cflxnODKgmk9dkyGP1P5KcqR45JPe7kvxy03hV7zDX16V35qS/XfqfNW8Sq5TyPaWUJx2qR5I8s5Tyzibx0tvWC1X/IytVVX1hemfw+FjDePckeXbV+6x4qqr6kvTOevLRMbHOVVV1x6HfX9DPo1F/jIj3QJP+GBHv5U36Y8z2Tt0fI+Ldlyn7Y1jPp3dE1tR9MSLeW9OgL0asSc9o0htjtnfq3hgR771p1hsbVVVt9v9/IcnXp9cTjdapYfFmWKcGxmu6Vo3Z3qnrMSLe1GvViNevd6XBOjXq9bDJOjUi3iubrFNjXq+nXqdGxJt13+rFSX6tlPKx/r8z637VdfH6Oc2yX3VjfrPuV90Yb5b9qhu3tWktht3v3WlWi6F5NKzFsHivaliLUY9Tk1oMi/eBEf/OSFVVPaH/czG9jxL/TCnlfWnYG0PiXWjaG0Pi/VjDfaph29qoLwbFS/KnmWGdGrUuNVmzxt1n2rqMye9I3wueJAvdbpOTn9CG/g7gn6S30/XwKZfvKaX89SOKf2+Sry4znK6xqqqnpXcI8W3pnX75H5VSfmOGeN+Y5B/kkS/ufVkp5U0T3vdfJPkbSZ6Q3s7tx0opz6yq6lPTO4PCren9pflFpf9X2GnjJanSsCbD8rvhNvdmwpqM2N5GNRkRr1FNRj1/q6r6/PTO5rCeR05T/qEm8dL7SMiv9q/b71/++6WU722S2w2362aCU5SP2daHD59dS+8v2X+/lPKHM8T7gSR/J72zZF1O70xZvzcm3uOTvDnJRnqP0QNJXlpKubtJfwyL189n6v4Yld8Nt7s3E/THmO2duj/GxGu8Zt24TU36Yli89I7WmaovxuV3w+UT9caoeE16Y0y8Jr3xtCT/Ib2P0ywleWeS7yulfLDhOjUwXnpfKjp1PUbld8PtJl2rRm1vk7VqVLym9bipP2d4Hb8pXnoD46av42PXjylfx4dtb9PX8WHxZtm3eld6Nf3NQ5c1qsegeLPu6w7K74br780U+7pDtrdpPQbFarpPNfB+M/TGTfHSG2Q17Y2x2zVlbwzb3qa1GBavaT1em97RnatJfivJS0opuzPU46Z46R0l2rQeA/O74Tb3ZrJ9qmHb2rQWw+JNVIsm7/fGXDdVvHFr1qzvR4/i/flJYfAEAAAAwLHwUTsAAAAAjoXBEwAAAADHwuAJAAAAgGNh8AQAAADAsTB4AgAAAOBYLLedAAAASVVV35Lk55LcVUp5z6HLn5vkLUnel+QrSikfbSdDAIDpOeIJAOCEqqrq85O8Ncm7k3ypoRMAMG8MngAATqCqqr4ovSOd/nd6Rzo92HJKAABTM3gCADhhqqr6iiS/keSPkjy/lFK3nBIAQCMGTwAAJ8tXJfnPSX4nyVeVUi60nA8AQGMGTwAAJ8tPJ/nzJF9TSrnUci4AADMxeAIAOFl+LcknJfnhthMBAJiVwRMAwMnykiT/NsnLqqr6obaTAQCYhcETAMDJ0k3yHUnekOTHq6r6/nbTAQBozuAJAOCEKaUcJHlxkl9O8oqqqr6r5ZQAABoxeAIAOIFKKftJviHJm5O8uqqqb205JQCAqRk8AQCcUKWUvSRVkt9M8tqqqr6h5ZQAAKay0O12284BAAAAgFPIEU8AAAAAHAuDJwAAAACOhcETAAAAAMfC4AkAAACAY2HwBAAAAMCxMHgCAAAA4FgYPAEAAABwLAyeAAAAADgW/x+LlM57ZgNmXwAAAABJRU5ErkJggg==\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"Cabe resaltar que conforme aumentamos a quantidade de features consideradas alguns indicios de overfitting começam a ser evidentes, isso é bem esperado dado que o aumento de features também aumenta a complexidade no modelo, permitindo modelos mais flexiveis que são mais propensos a overffitar. Com certeza este tipo de comportamento seria mais evidentes em outros tipos de modelos (resaltando que \"overffitar um modelo randomforest é quase que um desafio\" <- Palavras do Felipe; só que não rsrs)","metadata":{}},{"cell_type":"markdown","source":"Fazendo um tradeoff entre melhor preformance vs número de features o experimento indicado seria o de indice 22 (escolha arbitraria), proseguiremos nosso desenvolvimento com esta escolha. Vemos que o metodo é pedorozo que resultamos com uma diminuição do número de features consideradas e um modelo com melhor performance.","metadata":{}},{"cell_type":"code","source":"\n\nimportances = K_estimator[22].feature_importances_\n#importances = rnd_clf.feature_importances_\nindices = np.argsort(importances)\n\nfig, ax = plt.subplots(figsize = (20, 15))\nplt.title('Feature Importances')\nplt.barh(range(len(indices)), importances[indices], color='b', align='center')\nplt.yticks(range(len(indices)), [K_features[22][i] for i in indices])\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-09-03T16:00:32.294716Z","iopub.execute_input":"2021-09-03T16:00:32.295149Z","iopub.status.idle":"2021-09-03T16:00:33.088694Z","shell.execute_reply.started":"2021-09-03T16:00:32.295112Z","shell.execute_reply":"2021-09-03T16:00:33.087606Z"},"trusted":true},"execution_count":56,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1440x1080 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"Apresentando a importancia das variaveis observamos que ainda existem algumas que contribuim relativamente pouco ao aprendizado do modelo. ","metadata":{}},{"cell_type":"code","source":"print('treino:')\nshow_performance(K_estimator[22], X_train[K_features[22]], y_train)\n\nprint('test:')\nshow_performance(K_estimator[22], X_test[K_features[22]], y_test)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T16:01:31.399752Z","iopub.execute_input":"2021-09-03T16:01:31.400125Z","iopub.status.idle":"2021-09-03T16:01:32.173985Z","shell.execute_reply.started":"2021-09-03T16:01:31.400094Z","shell.execute_reply":"2021-09-03T16:01:32.173147Z"},"trusted":true},"execution_count":57,"outputs":[{"name":"stdout","text":"treino:\nMatriz de confusão:\n[[52641  5769]\n [  944  1462]]\n\nLucro obtido com a ação (em Reais): R$ 73890 \nLucro maximo possivel: R$ 216540 \nPorcentagem do lucro maximo: 34.12%\n\ntest:\nMatriz de confusão:\n[[13145  1457]\n [  262   340]]\n\nLucro obtido com a ação (em Reais): R$ 16030 \nLucro maximo possivel: R$ 54180 \nPorcentagem do lucro maximo: 29.59%\n\n","output_type":"stream"}]},{"cell_type":"code","source":"len(K_features[22])","metadata":{"execution":{"iopub.status.busy":"2021-09-03T18:59:52.872473Z","iopub.execute_input":"2021-09-03T18:59:52.872828Z","iopub.status.idle":"2021-09-03T18:59:52.878975Z","shell.execute_reply.started":"2021-09-03T18:59:52.872798Z","shell.execute_reply":"2021-09-03T18:59:52.877963Z"},"trusted":true},"execution_count":138,"outputs":[{"execution_count":138,"output_type":"execute_result","data":{"text/plain":"46"},"metadata":{}}]},{"cell_type":"code","source":"model = K_estimator[22]\nselected_features = K_features[22]","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:04:20.166491Z","iopub.execute_input":"2021-09-03T19:04:20.167006Z","iopub.status.idle":"2021-09-03T19:04:20.170945Z","shell.execute_reply.started":"2021-09-03T19:04:20.166974Z","shell.execute_reply":"2021-09-03T19:04:20.169929Z"},"trusted":true},"execution_count":146,"outputs":[]},{"cell_type":"code","source":"model.fit(X_train[selected_features], y_train)\n\nprint('treino:')\nshow_performance(model, X_train[selected_features], y_train)\n\nprint('test:')\nshow_performance(model, X_test[selected_features], y_test)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:04:21.454243Z","iopub.execute_input":"2021-09-03T19:04:21.454827Z","iopub.status.idle":"2021-09-03T19:04:23.395751Z","shell.execute_reply.started":"2021-09-03T19:04:21.454793Z","shell.execute_reply":"2021-09-03T19:04:23.394807Z"},"trusted":true},"execution_count":147,"outputs":[{"name":"stdout","text":"treino:\nMatriz de confusão:\n[[52641  5769]\n [  944  1462]]\n\nLucro obtido com a ação (em Reais): R$ 73890 \nLucro maximo possivel: R$ 216540 \nPorcentagem do lucro maximo: 34.12%\n\ntest:\nMatriz de confusão:\n[[13145  1457]\n [  262   340]]\n\nLucro obtido com a ação (em Reais): R$ 16030 \nLucro maximo possivel: R$ 54180 \nPorcentagem do lucro maximo: 29.59%\n\n","output_type":"stream"}]},{"cell_type":"markdown","source":"Deixaremos por aqui a parte 1, ficando pendente nos extras (se der tempo) alguns testes obrigatorios como são:\n  - Explorar variaveis categóricas\n  - Utilizar Underfitting e Overfitting\n  - Testar outros modelos XGboost e LightGBM\n  - mais feature engineering: combinação de features, criação de novas features, transformação de variaveis, normalização, tratamento dos outliers.\n  - Testar criando a variavel de combinação de feature com variancias proximas a zero.","metadata":{}},{"cell_type":"markdown","source":"# Parte 2","metadata":{}},{"cell_type":"markdown","source":"Nesta parte exploraremos segmentações dos clientes baseado na probabilidade de eles serem insatisfeitos. Estaprobabilidade é obtida a partir da função predit_proba()","metadata":{}},{"cell_type":"markdown","source":"Neste casso utilizaremos a metrica AUC para escolher o melhor tradeoff entre TP/FP, levando em conta que o custo de cada TP é diferente ao de um FP","metadata":{}},{"cell_type":"code","source":"from sklearn.metrics import roc_auc_score as auc","metadata":{"execution":{"iopub.status.busy":"2021-09-03T16:02:06.887130Z","iopub.execute_input":"2021-09-03T16:02:06.887695Z","iopub.status.idle":"2021-09-03T16:02:06.891318Z","shell.execute_reply.started":"2021-09-03T16:02:06.887659Z","shell.execute_reply":"2021-09-03T16:02:06.890599Z"},"trusted":true},"execution_count":59,"outputs":[]},{"cell_type":"code","source":"y_prob_test = model.predict_proba(X_test[selected_features])[:,1]\ny_prob_train = model.predict_proba(X_train[selected_features])[:,1]","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:04:34.667866Z","iopub.execute_input":"2021-09-03T19:04:34.668235Z","iopub.status.idle":"2021-09-03T19:04:35.003588Z","shell.execute_reply.started":"2021-09-03T19:04:34.668204Z","shell.execute_reply":"2021-09-03T19:04:35.002591Z"},"trusted":true},"execution_count":148,"outputs":[]},{"cell_type":"code","source":"fig, ax = plt.subplots(figsize = (18, 8))\nax.hist(y_prob_test, bins = 20);\nax.set_xlim(0, 1);\nplt.xticks(np.arange(0, 1, 0.1))\nplt.title('Probability distribution for unsatisfied classification', fontsize=18);\nplt.ylabel('Frequency', fontsize=16);\nplt.xlabel('Probability', fontsize=16);\nax.spines['top'].set_visible(False)\nax.spines['right'].set_visible(False)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T16:02:31.235882Z","iopub.execute_input":"2021-09-03T16:02:31.236266Z","iopub.status.idle":"2021-09-03T16:02:31.545693Z","shell.execute_reply.started":"2021-09-03T16:02:31.236232Z","shell.execute_reply":"2021-09-03T16:02:31.544600Z"},"trusted":true},"execution_count":61,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1296x576 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"a distribuição mostra que nosso modelo prediz como insatisfeitos aqueles cuja probabilidade seja > 0.5. Dado que este grupo já está bem definido pelo modelo, vamos respeitar este limite para criar nosso primeiro grupo, ou grupo objetivo da campanha de retenção.","metadata":{}},{"cell_type":"code","source":"# Code base on this post: https://stackoverflow.com/questions/25009284/how-to-plot-roc-curve-in-python\nimport sklearn.metrics as metrics\n\n# Calculate FPR and TPR for all thresholds\nfpr, tpr, threshold = metrics.roc_curve(y_train, y_prob_train)\nroc_auc = metrics.auc(fpr, tpr)\n\n# Plotting the ROC curve\nimport matplotlib.pyplot as plt\nfig, ax = plt.subplots(figsize = (20, 8))\nplt.title('Curva ROC', fontsize=18)\nplt.plot(fpr, tpr, 'b', label = 'AUC = %0.4f' % roc_auc)\nax.spines['top'].set_visible(False)\nax.spines['right'].set_visible(False)\nplt.legend(loc = 'upper left', fontsize = 16)\nplt.plot([0, 1], [0, 1],'r--')\nplt.xlim([0, 1])\nplt.ylim([0, 1])\nplt.ylabel('True Positive Rate', fontsize = 16)\nplt.xlabel('False Positive Rate', fontsize = 16)\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-09-03T16:02:56.596403Z","iopub.execute_input":"2021-09-03T16:02:56.596941Z","iopub.status.idle":"2021-09-03T16:02:56.846196Z","shell.execute_reply.started":"2021-09-03T16:02:56.596910Z","shell.execute_reply":"2021-09-03T16:02:56.845427Z"},"trusted":true},"execution_count":63,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1440x576 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"print('auc_train', auc(y_train, y_prob_train))\nprint('auc_test', auc(y_test, y_prob_test))","metadata":{"execution":{"iopub.status.busy":"2021-09-03T16:02:48.594827Z","iopub.execute_input":"2021-09-03T16:02:48.595212Z","iopub.status.idle":"2021-09-03T16:02:48.629849Z","shell.execute_reply.started":"2021-09-03T16:02:48.595182Z","shell.execute_reply":"2021-09-03T16:02:48.629039Z"},"trusted":true},"execution_count":62,"outputs":[{"name":"stdout","text":"auc_train 0.8773190824513788\nauc_test 0.844067519536076\n","output_type":"stream"}]},{"cell_type":"markdown","source":"A curva ROC é uma visualização poderoza de como nosso modelo está performando em função dos FP e FN. Ela é construida ranqueando os observações segundo as probabilidades calculadas. Conforme for diminuido o nível da probabilidade, são calculadas as frações de FP e FN que entram no nível. Ela serve usualmente para definir o melhor threshold para definir um publico objetivo baseado nas proporções de FP e FN que entram. Neste casso não utilizaremos a curva ROC dado que o custo entre FP e FN não são equivalentes. O ideal no nosso casso seria definir a função de lucro em função do threshold de probabilidade escolhido (conforme apresentado por CaioMar).","metadata":{}},{"cell_type":"code","source":"def segmentacao(x):\n    delta = 0.5/4\n    if x >= 0.5:\n        return 'grupo 1'\n    if x >= 0.5-delta:\n        return 'grupo 2'\n    if x >= 0.5-2*delta:\n        return 'grupo 3'\n    if x >= 0.5-3*delta:\n        return 'grupo 4'\n    if x >= 0.5-4*delta:\n        return 'grupo 5'\n    \n\ny_test_aux = y_test.copy()\n#y_prob_test\n\ny_prob_test_df = pd.DataFrame(data = y_prob_test, columns = ['y_prob_test'], index = y_test.index.copy())\ny_test_aux_out = pd.merge(y_test_aux, y_prob_test_df, how = 'left', left_index = True, right_index = True)\n\ny_test_aux_out['segmento'] = y_test_aux_out['y_prob_test'].apply(lambda x: segmentacao(x))\n\n\ny_test_aux_out.groupby([target,'segmento']).size().unstack(target).plot.bar(stacked=True,figsize = (16,6))","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:26:30.153485Z","iopub.execute_input":"2021-09-03T19:26:30.154108Z","iopub.status.idle":"2021-09-03T19:26:30.426598Z","shell.execute_reply.started":"2021-09-03T19:26:30.154068Z","shell.execute_reply":"2021-09-03T19:26:30.425645Z"},"trusted":true},"execution_count":171,"outputs":[{"execution_count":171,"output_type":"execute_result","data":{"text/plain":"<AxesSubplot:xlabel='segmento'>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<Figure size 1152x432 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"y_test_aux_out.groupby([target,'segmento']).size()","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:02:12.613699Z","iopub.execute_input":"2021-09-03T20:02:12.614072Z","iopub.status.idle":"2021-09-03T20:02:12.629509Z","shell.execute_reply.started":"2021-09-03T20:02:12.614041Z","shell.execute_reply":"2021-09-03T20:02:12.628368Z"},"trusted":true},"execution_count":185,"outputs":[{"execution_count":185,"output_type":"execute_result","data":{"text/plain":"TARGET  segmento\n0       grupo 1     1457\n        grupo 2     1073\n        grupo 3     2084\n        grupo 4     3785\n        grupo 5     6203\n1       grupo 1      340\n        grupo 2       79\n        grupo 3       87\n        grupo 4       71\n        grupo 5       25\ndtype: int64"},"metadata":{}}]},{"cell_type":"markdown","source":"Temos definido os grupos quase que manualmente uma vez sabendo como o nosso modelo está se comportando, no grupo 1 estamos separando os que o modelo tinha classificado como insatisfeitos. Neste casso sabemos que a performance é de 29.59% do ideal do grupo, com lucro de R\\$ 16030. Cabe uma escolha mais inteligente de como são divididos os grupos 2 a 5.","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:26:30.427993Z","iopub.execute_input":"2021-09-03T19:26:30.428269Z","iopub.status.idle":"2021-09-03T19:26:30.433852Z","shell.execute_reply.started":"2021-09-03T19:26:30.428242Z","shell.execute_reply":"2021-09-03T19:26:30.432706Z"}}},{"cell_type":"code","source":"340*90 - 1457*10","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:02:56.782971Z","iopub.execute_input":"2021-09-03T20:02:56.783540Z","iopub.status.idle":"2021-09-03T20:02:56.790440Z","shell.execute_reply.started":"2021-09-03T20:02:56.783507Z","shell.execute_reply":"2021-09-03T20:02:56.789112Z"},"trusted":true},"execution_count":186,"outputs":[{"execution_count":186,"output_type":"execute_result","data":{"text/plain":"16030"},"metadata":{}}]},{"cell_type":"markdown","source":"# Parte 3","metadata":{}},{"cell_type":"markdown","source":"Neste casso será utilizado o método de clustering K-means apos um tratamento para reescalar as features. O reescalamento dos dados é necesario devido a que K-means é sensivel a escala.\n\nTambem seria recomendado utilizar um metodo de redução de dimensionalidade (LDA, PCA, t-SNE) antes de aplicar o clustering, mas neste casso vamos com nossas features escolhidas na parte 1.\n\nAqui o método consiste de \n- 1. Reescalar as features, \n- 2. Iterar sobre um range escolhido de Ks, enquanto são sapturados os dados de numero de Ks e scores calculados, aqui aproveitamos também para calcular o score da metrica silhouette\n(lembrando que o score original no metodo K-means (também conhecido como inercia) é relacionado à somatoria de quadrados intraclusters, enquanto que o score silhouette leva em conta distancia intracluster e interclusters) \n- 3. Representação grafica dos resultados para escolher do melhor K","metadata":{}},{"cell_type":"code","source":"%%time\nfrom sklearn.cluster import KMeans\nfrom sklearn.preprocessing import MinMaxScaler\nfrom sklearn.metrics import silhouette_score\n\n\n\nscaler = MinMaxScaler()\n\nX_train_scale = pd.DataFrame(scaler.fit_transform(X_train[selected_features]), columns=selected_features)\nX_test_scale = pd.DataFrame(scaler.transform(X_test[selected_features]), columns=selected_features)\n\n\ninertials = []\nsilhouette_scores = []\n\nfor i in range(2, 50):\n    kmeans = KMeans(n_clusters = i, init = 'random')\n    kmeans.fit(X_train_scale)\n    inertials.append(kmeans.inertia_)\n    score = silhouette_score(X_train_scale, kmeans.labels_)\n    silhouette_scores.append(score)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T16:30:50.032694Z","iopub.execute_input":"2021-09-03T16:30:50.033107Z","iopub.status.idle":"2021-09-03T17:28:49.114196Z","shell.execute_reply.started":"2021-09-03T16:30:50.033073Z","shell.execute_reply":"2021-09-03T17:28:49.112882Z"},"trusted":true},"execution_count":83,"outputs":[{"name":"stdout","text":"CPU times: user 1h 28min 11s, sys: 27min 50s, total: 1h 56min 2s\nWall time: 57min 59s\n","output_type":"stream"}]},{"cell_type":"code","source":"# Elbow Curve\nfig, ax = plt.subplots(figsize = (20, 8))\nplt.plot(range(2, 50), inertials)\nplt.title('Elbow Curve', fontsize = 18)\nplt.xlabel('Number of Clusters', fontsize = 16)\nplt.xticks(np.arange(2, 50, 1))\nplt.ylabel('Within Cluster Sum of Squares', fontsize = 16) \nax.spines['top'].set_visible(False)\nax.spines['right'].set_visible(False)\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-09-03T17:28:49.115826Z","iopub.execute_input":"2021-09-03T17:28:49.116143Z","iopub.status.idle":"2021-09-03T17:28:49.779737Z","shell.execute_reply.started":"2021-09-03T17:28:49.116108Z","shell.execute_reply":"2021-09-03T17:28:49.778416Z"},"trusted":true},"execution_count":84,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1440x576 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"# Silhouette\nfig, ax = plt.subplots(figsize = (20, 8))\nplt.plot(range(2, 50), silhouette_scores)\nplt.title('Silhouette', fontsize = 18)\nplt.xlabel('Value of K', fontsize = 16)\nplt.xticks(np.arange(6, 50, 1))\nplt.ylabel('Silhouette Score', fontsize = 16) \nax.spines['top'].set_visible(False)\nax.spines['right'].set_visible(False)\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-09-03T17:29:23.231524Z","iopub.execute_input":"2021-09-03T17:29:23.232174Z","iopub.status.idle":"2021-09-03T17:29:23.838794Z","shell.execute_reply.started":"2021-09-03T17:29:23.232135Z","shell.execute_reply":"2021-09-03T17:29:23.837752Z"},"trusted":true},"execution_count":86,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1440x576 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"Pelo metodo Silhouette, é evidente que o melhor número de cluster é 13 ou 17 apresentando score aproximados. Ficaremos com 13 pela facilidade de ter menor numero de clusters.","metadata":{}},{"cell_type":"code","source":"kmeans = KMeans(n_clusters = 17).fit(X_train_scale)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T17:39:11.425756Z","iopub.execute_input":"2021-09-03T17:39:11.426322Z","iopub.status.idle":"2021-09-03T17:39:17.591854Z","shell.execute_reply.started":"2021-09-03T17:39:11.426275Z","shell.execute_reply":"2021-09-03T17:39:17.590284Z"},"trusted":true},"execution_count":102,"outputs":[]},{"cell_type":"code","source":"result_kmean_train = pd.DataFrame({'target': y_train, 'labels': kmeans.labels_})\nunsatisfied_dist = result_kmean_train[result_kmean_train['target'] == 1].labels.value_counts().sort_index()\n\n# Plotting the unsatified customers ditribution for the clusters\nfig, ax = plt.subplots(figsize = (20, 8))\nplt.bar(unsatisfied_dist.index + 1, unsatisfied_dist.values);\nplt.title('Unsatisfied customer amount for each cluster', fontsize = 18);\nplt.xlabel('Cluster', fontsize = 16);\nplt.ylabel('Amount', fontsize = 16)\nplt.xticks(range(1, 26, 1))\n\nax.spines['top'].set_visible(False)\nax.spines['right'].set_visible(False)\nax.spines['left'].set_visible(False)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T17:39:19.617280Z","iopub.execute_input":"2021-09-03T17:39:19.617700Z","iopub.status.idle":"2021-09-03T17:39:20.070930Z","shell.execute_reply.started":"2021-09-03T17:39:19.617667Z","shell.execute_reply":"2021-09-03T17:39:20.069862Z"},"trusted":true},"execution_count":103,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1440x576 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"resumo = pd.crosstab(result_kmean_train['labels'],result_kmean_train['target'])\nresumo['lucro'] = resumo[1]*90 - resumo[0]*10\nresumo","metadata":{"execution":{"iopub.status.busy":"2021-09-03T17:39:23.666580Z","iopub.execute_input":"2021-09-03T17:39:23.666969Z","iopub.status.idle":"2021-09-03T17:39:23.698880Z","shell.execute_reply.started":"2021-09-03T17:39:23.666937Z","shell.execute_reply":"2021-09-03T17:39:23.697901Z"},"trusted":true},"execution_count":104,"outputs":[{"execution_count":104,"output_type":"execute_result","data":{"text/plain":"target      0     1   lucro\nlabels                     \n0        7522   133  -63250\n1       10700  1015  -15650\n2        2652   101  -17430\n3        1107    14   -9810\n4        5335   152  -39670\n5        1400    21  -12110\n6        1172    95   -3170\n7        3338   364    -620\n8        1240    10  -11500\n9       17502   351 -143430\n10       1073     8  -10010\n11        717    37   -3840\n12       2451    43  -20640\n13        226     7   -1630\n14        740     4   -7040\n15        780    17   -6270\n16        455    34   -1490","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th>target</th>\n      <th>0</th>\n      <th>1</th>\n      <th>lucro</th>\n    </tr>\n    <tr>\n      <th>labels</th>\n      <th></th>\n      <th></th>\n      <th></th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>7522</td>\n      <td>133</td>\n      <td>-63250</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>10700</td>\n      <td>1015</td>\n      <td>-15650</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>2652</td>\n      <td>101</td>\n      <td>-17430</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>1107</td>\n      <td>14</td>\n      <td>-9810</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>5335</td>\n      <td>152</td>\n      <td>-39670</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>1400</td>\n      <td>21</td>\n      <td>-12110</td>\n    </tr>\n    <tr>\n      <th>6</th>\n      <td>1172</td>\n      <td>95</td>\n      <td>-3170</td>\n    </tr>\n    <tr>\n      <th>7</th>\n      <td>3338</td>\n      <td>364</td>\n      <td>-620</td>\n    </tr>\n    <tr>\n      <th>8</th>\n      <td>1240</td>\n      <td>10</td>\n      <td>-11500</td>\n    </tr>\n    <tr>\n      <th>9</th>\n      <td>17502</td>\n      <td>351</td>\n      <td>-143430</td>\n    </tr>\n    <tr>\n      <th>10</th>\n      <td>1073</td>\n      <td>8</td>\n      <td>-10010</td>\n    </tr>\n    <tr>\n      <th>11</th>\n      <td>717</td>\n      <td>37</td>\n      <td>-3840</td>\n    </tr>\n    <tr>\n      <th>12</th>\n      <td>2451</td>\n      <td>43</td>\n      <td>-20640</td>\n    </tr>\n    <tr>\n      <th>13</th>\n      <td>226</td>\n      <td>7</td>\n      <td>-1630</td>\n    </tr>\n    <tr>\n      <th>14</th>\n      <td>740</td>\n      <td>4</td>\n      <td>-7040</td>\n    </tr>\n    <tr>\n      <th>15</th>\n      <td>780</td>\n      <td>17</td>\n      <td>-6270</td>\n    </tr>\n    <tr>\n      <th>16</th>\n      <td>455</td>\n      <td>34</td>\n      <td>-1490</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"# Get the distribution only for unsatisfied customer on X_test_selected\nresult_kmean_test = pd.DataFrame({'target': y_test, 'labels': kmeans.predict(X_test_scale)})\nunsatisfied_dist_test = result_kmean_test[result_kmean_test['target'] == 1].labels.value_counts().sort_index()\n\n# Plotting the unsatified customers ditribution for the clusters\nfig, ax = plt.subplots(figsize = (20, 8))\nplt.bar(unsatisfied_dist_test.index + 1, unsatisfied_dist_test.values);\nplt.title('Unsatisfied customer amount for each cluster', fontsize = 18);\nplt.xlabel('Cluster', fontsize = 16);\nplt.ylabel('Amount', fontsize = 16)\nplt.xticks(range(1, 26, 1))\n\nax.spines['top'].set_visible(False)\nax.spines['right'].set_visible(False)\nax.spines['left'].set_visible(False)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T17:30:53.476589Z","iopub.execute_input":"2021-09-03T17:30:53.477126Z","iopub.status.idle":"2021-09-03T17:30:54.006881Z","shell.execute_reply.started":"2021-09-03T17:30:53.477094Z","shell.execute_reply":"2021-09-03T17:30:54.006028Z"},"trusted":true},"execution_count":92,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1440x576 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"resumo_test = pd.crosstab(result_kmean_test['labels'],result_kmean_test['target'])\nresumo_test['lucro'] = resumo_test[1]*90 - resumo_test[0]*10\nresumo_test","metadata":{"execution":{"iopub.status.busy":"2021-09-03T17:40:17.946357Z","iopub.execute_input":"2021-09-03T17:40:17.946927Z","iopub.status.idle":"2021-09-03T17:40:17.981340Z","shell.execute_reply.started":"2021-09-03T17:40:17.946890Z","shell.execute_reply":"2021-09-03T17:40:17.980394Z"},"trusted":true},"execution_count":105,"outputs":[{"execution_count":105,"output_type":"execute_result","data":{"text/plain":"target     0    1  lucro\nlabels                  \n0        318    3  -2910\n1       5035  110 -40450\n2       2741  267  -3380\n3        304    0  -3040\n4        852   87   -690\n5        387   31  -1080\n6       1376   41 -10070\n7        168    4  -1320\n8       1891   34 -15850\n9        598    9  -5170\n10       323    3  -2960\n11       172   10   -820\n12       437    3  -4100","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th>target</th>\n      <th>0</th>\n      <th>1</th>\n      <th>lucro</th>\n    </tr>\n    <tr>\n      <th>labels</th>\n      <th></th>\n      <th></th>\n      <th></th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>318</td>\n      <td>3</td>\n      <td>-2910</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>5035</td>\n      <td>110</td>\n      <td>-40450</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>2741</td>\n      <td>267</td>\n      <td>-3380</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>304</td>\n      <td>0</td>\n      <td>-3040</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>852</td>\n      <td>87</td>\n      <td>-690</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>387</td>\n      <td>31</td>\n      <td>-1080</td>\n    </tr>\n    <tr>\n      <th>6</th>\n      <td>1376</td>\n      <td>41</td>\n      <td>-10070</td>\n    </tr>\n    <tr>\n      <th>7</th>\n      <td>168</td>\n      <td>4</td>\n      <td>-1320</td>\n    </tr>\n    <tr>\n      <th>8</th>\n      <td>1891</td>\n      <td>34</td>\n      <td>-15850</td>\n    </tr>\n    <tr>\n      <th>9</th>\n      <td>598</td>\n      <td>9</td>\n      <td>-5170</td>\n    </tr>\n    <tr>\n      <th>10</th>\n      <td>323</td>\n      <td>3</td>\n      <td>-2960</td>\n    </tr>\n    <tr>\n      <th>11</th>\n      <td>172</td>\n      <td>10</td>\n      <td>-820</td>\n    </tr>\n    <tr>\n      <th>12</th>\n      <td>437</td>\n      <td>3</td>\n      <td>-4100</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"markdown","source":"Com o conjunto de features escolhidas não observamos que o método de clustering crie um cluster que concentre os clientes insatistefitos. Infelizmente fracazamos nesta tarefa, tentemos uma redução de dimensionalidade antes do agrupamento.","metadata":{}},{"cell_type":"code","source":"X_all = data[all_features] \ny_all = data[target]\n\nX_train, X_test, y_train, y_test = train_test_split(X_all, y_all, stratify = y_all, test_size = 0.20 ,random_state = 1234)\n\nprint(\"Training set has {} samples and {} features.\".format(X_train.shape[0], X_train.shape[1]))\nprint(\"Testing set has {} samples and {} features.\".format(X_test.shape[0], X_test.shape[1]))\nprint(\"Train target {:0.5f}%\".format(y_train.sum()/y_train.shape[0]))\nprint(\"Test target {:0.5f}%\".format(y_test.sum()/y_test.shape[0]))\n\n\nselected_features = list(\n                        set(X_train.columns) \n                        - set(zero_variance) \n                        - set(duplicate_vars)\n                        - set(features_nzv) \n                        - set(hight_corr)\n                        )\nlen(selected_features)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"n_comp=12\npca = PCA(n_components=n_comp)\n\nX_train_pca = pca.fit_transform(X_train_scale)\nX_test_pca = pca.transform(X_train_scale)\n\nprint(sum(pca.explained_variance_ratio_))","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"inertials_2 = [] \nsilhouette_scores_2 = []\nfor i in range(3, 25):\n    kmeans = KMeans(n_clusters = i, init = 'random')\n    kmeans.fit(X_train_pca)\n    inertials_2.append(kmeans.inertia_)\n    score = silhouette_score(X_train_pca, kmeans.labels_)\n    silhouette_scores_2.append(score)\n    print(i)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:28:16.362238Z","iopub.execute_input":"2021-09-03T19:28:16.362857Z","iopub.status.idle":"2021-09-03T19:54:44.512273Z","shell.execute_reply.started":"2021-09-03T19:28:16.362820Z","shell.execute_reply":"2021-09-03T19:54:44.510762Z"},"trusted":true},"execution_count":175,"outputs":[{"name":"stdout","text":"3\n4\n5\n6\n7\n8\n9\n10\n11\n12\n13\n14\n15\n16\n17\n18\n19\n20\n21\n22\n23\n24\n","output_type":"stream"}]},{"cell_type":"code","source":"# Silhouette\nfig, ax = plt.subplots(figsize = (20, 8))\nplt.plot(range(3, 25), silhouette_scores_2)\nplt.title('Silhouette', fontsize = 18)\nplt.xlabel('Value of K', fontsize = 16)\nplt.xticks(np.arange(3, 25))\nplt.ylabel('Silhouette Score', fontsize = 16) \nax.spines['top'].set_visible(False)\nax.spines['right'].set_visible(False)\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:56:01.624550Z","iopub.execute_input":"2021-09-03T19:56:01.625063Z","iopub.status.idle":"2021-09-03T19:56:01.961634Z","shell.execute_reply.started":"2021-09-03T19:56:01.625018Z","shell.execute_reply":"2021-09-03T19:56:01.960906Z"},"trusted":true},"execution_count":178,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1440x576 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"kmeans_2 = KMeans(n_clusters = 11).fit(X_train_pca)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:57:04.874304Z","iopub.execute_input":"2021-09-03T19:57:04.874664Z","iopub.status.idle":"2021-09-03T19:57:08.533278Z","shell.execute_reply.started":"2021-09-03T19:57:04.874635Z","shell.execute_reply":"2021-09-03T19:57:08.532451Z"},"trusted":true},"execution_count":179,"outputs":[]},{"cell_type":"code","source":"result_kmean_train_2 = pd.DataFrame({'target': y_train, 'labels': kmeans_2.labels_})\n#unsatisfied_dist_2 = result_kmean_train_2[result_kmean_train_2['target'] == 1].labels.value_counts().sort_index()\n","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:58:15.250639Z","iopub.execute_input":"2021-09-03T19:58:15.251056Z","iopub.status.idle":"2021-09-03T19:58:15.256934Z","shell.execute_reply.started":"2021-09-03T19:58:15.251024Z","shell.execute_reply":"2021-09-03T19:58:15.255849Z"},"trusted":true},"execution_count":180,"outputs":[]},{"cell_type":"code","source":"resumo_2 = pd.crosstab(result_kmean_train_2['labels'],result_kmean_train_2['target'])\nresumo_2['lucro'] = resumo_2[1]*90 - resumo_2[0]*10\nresumo_2","metadata":{"execution":{"iopub.status.busy":"2021-09-03T19:59:05.682386Z","iopub.execute_input":"2021-09-03T19:59:05.682904Z","iopub.status.idle":"2021-09-03T19:59:05.716456Z","shell.execute_reply.started":"2021-09-03T19:59:05.682873Z","shell.execute_reply":"2021-09-03T19:59:05.715412Z"},"trusted":true},"execution_count":181,"outputs":[{"execution_count":181,"output_type":"execute_result","data":{"text/plain":"target      0     1   lucro\nlabels                     \n0       10700  1015  -15650\n1       20275   453 -161980\n2        1623   127   -4800\n3        7543   136  -63190\n4        2460    28  -22080\n5        5401   152  -40330\n6        3287    60  -27470\n7        3340   364    -640\n8        1813    12  -17050\n9        1238    19  -10670\n10        730    40   -3700","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th>target</th>\n      <th>0</th>\n      <th>1</th>\n      <th>lucro</th>\n    </tr>\n    <tr>\n      <th>labels</th>\n      <th></th>\n      <th></th>\n      <th></th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>10700</td>\n      <td>1015</td>\n      <td>-15650</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>20275</td>\n      <td>453</td>\n      <td>-161980</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>1623</td>\n      <td>127</td>\n      <td>-4800</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>7543</td>\n      <td>136</td>\n      <td>-63190</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>2460</td>\n      <td>28</td>\n      <td>-22080</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>5401</td>\n      <td>152</td>\n      <td>-40330</td>\n    </tr>\n    <tr>\n      <th>6</th>\n      <td>3287</td>\n      <td>60</td>\n      <td>-27470</td>\n    </tr>\n    <tr>\n      <th>7</th>\n      <td>3340</td>\n      <td>364</td>\n      <td>-640</td>\n    </tr>\n    <tr>\n      <th>8</th>\n      <td>1813</td>\n      <td>12</td>\n      <td>-17050</td>\n    </tr>\n    <tr>\n      <th>9</th>\n      <td>1238</td>\n      <td>19</td>\n      <td>-10670</td>\n    </tr>\n    <tr>\n      <th>10</th>\n      <td>730</td>\n      <td>40</td>\n      <td>-3700</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"X_train_scale = pd.DataFrame(scaler.fit_transform(X_train[selected_features]), columns=selected_features)\nX_test_scale = pd.DataFrame(scaler.transform(X_test[selected_features]), columns=selected_features)\n\n\ninertials = []\nsilhouette_scores = []\n\nfor i in range(2, 50):\n    kmeans = KMeans(n_clusters = i, init = 'random')\n    kmeans.fit(X_train_scale)\n    inertials.append(kmeans.inertia_)\n    score = silhouette_score(X_train_scale, kmeans.labels_)\n    silhouette_scores.append(score)","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Extra","metadata":{}},{"cell_type":"code","source":"def limiar(x, l=0.5):\n    if x >= l:\n        return 1\n    else:\n        return 0\n\nthr_test = y_test.to_frame()\nthr_test['predicted'] = y_predicted\nthr_test['predicted_t'] = thr_test['predicted'].apply(lambda x: limiar(x, 0.5))\nmax_profit_score(thr_test.TARGET, thr_test.predicted_t)\n\n","metadata":{"execution":{"iopub.status.busy":"2021-09-01T22:49:46.170674Z","iopub.execute_input":"2021-09-01T22:49:46.171235Z","iopub.status.idle":"2021-09-01T22:49:46.226455Z","shell.execute_reply.started":"2021-09-01T22:49:46.171166Z","shell.execute_reply":"2021-09-01T22:49:46.224873Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Test com tratamento de variaveis categoricas","metadata":{}},{"cell_type":"code","source":"qtd = get_uniques(X_train[selected_features])\ncat_features = list(qtd[(qtd['qtd_unique_values'] >= 3) & (qtd['qtd_unique_values'] <= 10)]['var'])\nqtd[(qtd['qtd_unique_values'] >= 3) & (qtd['qtd_unique_values'] <= 10)].reset_index(drop=True)\n\nfor i in cat_features:\n    print(i, X_train[i].unique())\n    \n\nX_v2 = X_all[selected_features].copy()\n\n\ntype_dummies = pd.get_dummies(X_v2[cat_features], columns=cat_features, drop_first=True)\n#type_dummies\nnew_X_all = pd.concat([X_v2.drop(cat_features,axis=1 ),\n                      type_dummies],\n                      axis=1)\nprint(new_X_all.shape, type_dummies.shape)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:07:14.680066Z","iopub.execute_input":"2021-09-03T20:07:14.680545Z","iopub.status.idle":"2021-09-03T20:07:14.896992Z","shell.execute_reply.started":"2021-09-03T20:07:14.680489Z","shell.execute_reply":"2021-09-03T20:07:14.895717Z"},"trusted":true},"execution_count":187,"outputs":[{"name":"stdout","text":"num_var13_0 [ 0  3  6  9 12 18 15]\nnum_var12 [0 3 6]\nnum_var41_0 [ 3  0  6  9 12 33 18 21]\nnum_var26_0 [ 0  3  9  6 33 15 12]\nnum_var30_0 [  3   9   6   0  12  15  24 114  21  18]\nnum_var13_corto [0 3 6]\nnum_aport_var13_hace3 [ 0  6  3 12  9 18]\nnum_var24 [0 3 6]\nnum_var13_largo [ 0  6  3  9 12 18 15]\nnum_var12_0 [  0   3   6   9 111]\nvar36 [99  2  3  1  0]\nnum_var30 [ 3  0  6  9 12 15 21 18]\nnum_meses_var8_ult3 [0 1 2 3]\nnum_var42 [ 3  0  6  9 12]\nnum_var5_0 [3 0 6 9]\nnum_meses_var5_ult3 [1 3 0 2]\n(76020, 115) (76020, 85)\n","output_type":"stream"}]},{"cell_type":"code","source":"type_dummies.shape","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:07:23.999570Z","iopub.execute_input":"2021-09-03T20:07:23.999993Z","iopub.status.idle":"2021-09-03T20:07:24.007332Z","shell.execute_reply.started":"2021-09-03T20:07:23.999959Z","shell.execute_reply":"2021-09-03T20:07:24.006175Z"},"trusted":true},"execution_count":188,"outputs":[{"execution_count":188,"output_type":"execute_result","data":{"text/plain":"(76020, 85)"},"metadata":{}}]},{"cell_type":"code","source":"X_train, X_test, y_train, y_test = train_test_split(new_X_all, y_all, stratify = y_all, test_size = 0.20 ,random_state = 1234)\n\nprint(\"Training set has {} samples and {} features.\".format(X_train.shape[0], X_train.shape[1]))\nprint(\"Testing set has {} samples and {} features.\".format(X_test.shape[0], X_test.shape[1]))\nprint(\"Train target {:0.5f}%\".format(y_train.sum()/y_train.shape[0]))\nprint(\"Test target {:0.5f}%\".format(y_test.sum()/y_test.shape[0]))","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:08:19.498484Z","iopub.execute_input":"2021-09-03T20:08:19.498891Z","iopub.status.idle":"2021-09-03T20:08:19.630060Z","shell.execute_reply.started":"2021-09-03T20:08:19.498859Z","shell.execute_reply":"2021-09-03T20:08:19.628830Z"},"trusted":true},"execution_count":189,"outputs":[{"name":"stdout","text":"Training set has 60816 samples and 115 features.\nTesting set has 15204 samples and 115 features.\nTrain target 0.03956%\nTest target 0.03959%\n","output_type":"stream"}]},{"cell_type":"code","source":"len(X_train.columns)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:08:21.412310Z","iopub.execute_input":"2021-09-03T20:08:21.412759Z","iopub.status.idle":"2021-09-03T20:08:21.421879Z","shell.execute_reply.started":"2021-09-03T20:08:21.412722Z","shell.execute_reply":"2021-09-03T20:08:21.420756Z"},"trusted":true},"execution_count":190,"outputs":[{"execution_count":190,"output_type":"execute_result","data":{"text/plain":"115"},"metadata":{}}]},{"cell_type":"code","source":"    gsearch = GridSearchCV(estimator=RandomForestClassifier(random_state=1234, n_jobs=-1), \n                           param_grid = {'max_depth':[18],\n                                         'n_estimators':[47],\n                                         'criterion': ['gini'],\n                                         'class_weight':[{0:1,1:12}, {0:1,1:13}, {0:1,1:14}], \n                                         'min_samples_split':[int(X_train.shape[0]*0.01),\n                                                              int(X_train.shape[0]*0.005)],\n                                         'max_features':['auto','log2',40]\n                                         }, \n                           scoring=max_profit_scorer, \n                           cv=skf.split(X_train_selected_fc, y_train), \n                           n_jobs=-1, \n                           verbose=0)\n    \n    gsearch.fit(X_train_selected_fc, y_train)\n    bst_max_profit_model = gsearch.best_estimator_\n    K_estimator.append(bst_max_profit_model)\n    \n    y_train_pred = bst_max_profit_model.predict(X_train_selected_fc)\n    y_test_pred = bst_max_profit_model.predict(selector_fc.transform(X_test_clean))\n\n    y_train_optime = max_profit_score(y_train, y_train)\n    y_test_optime = max_profit_score(y_test, y_test)\n    \n    K_vs_score_fc_train.append(100*max_profit_score(y_train, y_train_pred)/y_train_optime)\n    K_vs_score_fc_test.append(100*max_profit_score(y_test, y_test_pred)/y_test_optime)\n    \n\n    end = time()\n    # Returning the metrics related to the tested K and the time spent on this iteration of the loop\n    print(\"index {}: -  k = {} - lucro_train = {} - lucro_test = {} - Time = {}s\".format(len(K_vs_score_fc_train)-1, k, K_vs_score_fc_train[-1], K_vs_score_fc_test[-1], end-start))","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"%%time\nparam_grid = {'max_depth':[18],\n              'n_estimators':[47],\n              'criterion': ['gini'],\n              'class_weight':[{0:1,1:12},\n                              {0:1,1:13},\n                              {0:1,1:14}], \n              'min_samples_split':[int(X_train.shape[0]*0.01),\n                                   int(X_train.shape[0]*0.005)],\n              'max_features':['auto','log2',40]\n             }\n\n\nskf = StratifiedKFold(n_splits=5, shuffle = True, random_state = 1234)\n\ngsearch = GridSearchCV(RandomForestClassifier(random_state=1234, n_jobs=-1), \n                       param_grid={'max_depth':[18],\n                                   'n_estimators':[47],\n                                   'criterion': ['gini'],\n                                   'class_weight':[{0:1,1:12},\n                                                   {0:1,1:13},\n                                                   {0:1,1:14}], \n                                   'min_samples_split':[int(X_train.shape[0]*0.01),\n                                                        int(X_train.shape[0]*0.005)],\n                                   'max_features':['auto','log2',40]\n                                  }, \n                       scoring=max_profit_scorer, \n                       cv=skf.split(X_train, y_train), \n                       n_jobs=-1, \n                       verbose=5)\n\ngsearch.fit(X_train, y_train)\n\n\nbst_max_profit_model = gsearch.best_estimator_\nprint('treino:')\nshow_performance(bst_max_profit_model, X_train, y_train)\n\nprint('test:')\nshow_performance(bst_max_profit_model, X_test, y_test)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:12:53.306393Z","iopub.execute_input":"2021-09-03T20:12:53.306839Z","iopub.status.idle":"2021-09-03T20:15:23.015950Z","shell.execute_reply.started":"2021-09-03T20:12:53.306806Z","shell.execute_reply":"2021-09-03T20:15:23.014847Z"},"trusted":true},"execution_count":192,"outputs":[{"name":"stdout","text":"Fitting 5 folds for each of 18 candidates, totalling 90 fits\n","output_type":"stream"},{"name":"stderr","text":"[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.\n[Parallel(n_jobs=-1)]: Done  10 tasks      | elapsed:   16.4s\n[Parallel(n_jobs=-1)]: Done  64 tasks      | elapsed:  1.7min\n[Parallel(n_jobs=-1)]: Done  90 out of  90 | elapsed:  2.4min finished\n","output_type":"stream"},{"name":"stdout","text":"treino:\nMatriz de confusão:\n[[51921  6489]\n [  814  1592]]\n\nLucro obtido com a ação (em Reais): R$ 78390 \nLucro maximo possivel: R$ 216540 \nPorcentagem do lucro maximo: 36.20%\n\ntest:\nMatriz de confusão:\n[[12934  1668]\n [  250   352]]\n\nLucro obtido com a ação (em Reais): R$ 15000 \nLucro maximo possivel: R$ 54180 \nPorcentagem do lucro maximo: 27.69%\n\nCPU times: user 14.1 s, sys: 278 ms, total: 14.4 s\nWall time: 2min 29s\n","output_type":"stream"}]},{"cell_type":"code","source":"%%time\n\nnew_X_all['random'] = np.random.randint(20, size=new_X_all.shape[0])\n\nX_train, X_test, y_train, y_test = train_test_split(new_X_all, y_all, stratify = y_all, test_size = 0.20 ,random_state = 1234)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:19:48.815914Z","iopub.execute_input":"2021-09-03T20:19:48.816373Z","iopub.status.idle":"2021-09-03T20:19:48.946020Z","shell.execute_reply.started":"2021-09-03T20:19:48.816331Z","shell.execute_reply":"2021-09-03T20:19:48.944862Z"},"trusted":true},"execution_count":193,"outputs":[{"name":"stdout","text":"CPU times: user 120 ms, sys: 4.54 ms, total: 124 ms\nWall time: 123 ms\n","output_type":"stream"}]},{"cell_type":"code","source":"print(\"Training set has {} samples and {} features.\".format(X_train.shape[0], X_train.shape[1]))\nprint(\"Testing set has {} samples and {} features.\".format(X_test.shape[0], X_test.shape[1]))\nprint(\"Train target {:0.5f}%\".format(y_train.sum()/y_train.shape[0]))\nprint(\"Test target {:0.5f}%\".format(y_test.sum()/y_test.shape[0]))\n\nparam_grid = {'max_depth':[18],\n              'n_estimators':[47],\n              'criterion': ['gini'],\n              'class_weight':[{0:1,1:12},\n                              {0:1,1:13},\n                              {0:1,1:14}], \n              'min_samples_split':[int(X_train.shape[0]*0.01),\n                                   int(X_train.shape[0]*0.005)],\n              'max_features':['auto','log2',40]\n             }\n\n\nskf = StratifiedKFold(n_splits=5, shuffle = True, random_state = 1234)\n\ngsearch = GridSearchCV(RandomForestClassifier(random_state=1234, n_jobs=-1), \n                       param_grid=param_grid, \n                       scoring=max_profit_scorer, \n                       cv=skf.split(X_train, y_train), \n                       n_jobs=-1, \n                       verbose=5)\n\ngsearch.fit(X_train, y_train)\n\n\nbst_max_profit_model = gsearch.best_estimator_\nprint('treino:')\nshow_performance(bst_max_profit_model, X_train, y_train)\n\nprint('test:')\nshow_performance(bst_max_profit_model, X_test, y_test)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:20:08.354844Z","iopub.execute_input":"2021-09-03T20:20:08.355190Z","iopub.status.idle":"2021-09-03T20:22:38.641185Z","shell.execute_reply.started":"2021-09-03T20:20:08.355161Z","shell.execute_reply":"2021-09-03T20:22:38.640441Z"},"trusted":true},"execution_count":195,"outputs":[{"name":"stdout","text":"Training set has 60816 samples and 116 features.\nTesting set has 15204 samples and 116 features.\nTrain target 0.03956%\nTest target 0.03959%\nFitting 5 folds for each of 18 candidates, totalling 90 fits\n","output_type":"stream"},{"name":"stderr","text":"[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.\n[Parallel(n_jobs=-1)]: Done  10 tasks      | elapsed:   14.2s\n[Parallel(n_jobs=-1)]: Done  64 tasks      | elapsed:  1.7min\n[Parallel(n_jobs=-1)]: Done  90 out of  90 | elapsed:  2.4min finished\n","output_type":"stream"},{"name":"stdout","text":"treino:\nMatriz de confusão:\n[[52399  6011]\n [  867  1539]]\n\nLucro obtido com a ação (em Reais): R$ 78400 \nLucro maximo possivel: R$ 216540 \nPorcentagem do lucro maximo: 36.21%\n\ntest:\nMatriz de confusão:\n[[13070  1532]\n [  261   341]]\n\nLucro obtido com a ação (em Reais): R$ 15370 \nLucro maximo possivel: R$ 54180 \nPorcentagem do lucro maximo: 28.37%\n\n","output_type":"stream"}]},{"cell_type":"code","source":"feat_imp = pd.Series(bst_max_profit_model.feature_importances_, index=X_train.columns)\nfeat_imp.sort_values(inplace=True)\nax = feat_imp.tail(50).plot(kind='barh', figsize=(15,15), title='Feature importance')","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:22:38.642496Z","iopub.execute_input":"2021-09-03T20:22:38.642893Z","iopub.status.idle":"2021-09-03T20:22:39.612263Z","shell.execute_reply.started":"2021-09-03T20:22:38.642862Z","shell.execute_reply":"2021-09-03T20:22:39.611208Z"},"trusted":true},"execution_count":196,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1080x1080 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"O resultado anterior nos mostra que, das variaveis categoricas não existe alguma que resulte com importancia melhor que a variavel random. Inacreditavel!!","metadata":{}},{"cell_type":"code","source":"feat_imp.sort_values(ascending= False, inplace=True)\n\nfeat_imp_df = pd.DataFrame(feat_imp).reset_index().reset_index()\nfeat_imp_df.columns = ['pos', 'var', 'imp']\nimportant = feat_imp_df[feat_imp_df['pos'] < int(feat_imp_df[feat_imp_df['var'] == 'random']['pos'])]\nimportant","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:22:39.613889Z","iopub.execute_input":"2021-09-03T20:22:39.614184Z","iopub.status.idle":"2021-09-03T20:22:39.635328Z","shell.execute_reply.started":"2021-09-03T20:22:39.614154Z","shell.execute_reply":"2021-09-03T20:22:39.634270Z"},"trusted":true},"execution_count":197,"outputs":[{"execution_count":197,"output_type":"execute_result","data":{"text/plain":"   pos                     var       imp\n0    0                   var15  0.322114\n1    1             saldo_var30  0.194644\n2    2                   var38  0.096660\n3    3  saldo_medio_var5_hace2  0.065290\n4    4   saldo_medio_var5_ult3  0.054800\n5    5               num_var35  0.039771\n6    6               ind_var30  0.038580","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>pos</th>\n      <th>var</th>\n      <th>imp</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>var15</td>\n      <td>0.322114</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>1</td>\n      <td>saldo_var30</td>\n      <td>0.194644</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>2</td>\n      <td>var38</td>\n      <td>0.096660</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>3</td>\n      <td>saldo_medio_var5_hace2</td>\n      <td>0.065290</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>4</td>\n      <td>saldo_medio_var5_ult3</td>\n      <td>0.054800</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>5</td>\n      <td>num_var35</td>\n      <td>0.039771</td>\n    </tr>\n    <tr>\n      <th>6</th>\n      <td>6</td>\n      <td>ind_var30</td>\n      <td>0.038580</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"markdown","source":"# Undersampling e Oversampling","metadata":{}},{"cell_type":"markdown","source":"Para os testes de OverSampler e Undersampler utilizamos a biblioteca imblearn a qual possui a implementação dos metodos RandomOverSampler, RandomUnderSampler","metadata":{}},{"cell_type":"code","source":"X_all = data[all_features] \ny_all = data[target]\n\nX_train, X_test, y_train, y_test = train_test_split(X_all, y_all, stratify = y_all, test_size = 0.20 ,random_state = 1234)\n\nprint(\"Training set has {} samples and {} features.\".format(X_train.shape[0], X_train.shape[1]))\nprint(\"Testing set has {} samples and {} features.\".format(X_test.shape[0], X_test.shape[1]))\nprint(\"Train target {:0.5f}%\".format(y_train.sum()/y_train.shape[0]))\nprint(\"Test target {:0.5f}%\".format(y_test.sum()/y_test.shape[0]))\n\n\nselected_features = list(\n                        set(X_train.columns) \n                        - set(zero_variance) \n                        - set(duplicate_vars)\n                        - set(features_nzv) \n                        - set(hight_corr)\n                        )\nlen(selected_features)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:25:54.966544Z","iopub.execute_input":"2021-09-03T20:25:54.966968Z","iopub.status.idle":"2021-09-03T20:25:55.548633Z","shell.execute_reply.started":"2021-09-03T20:25:54.966934Z","shell.execute_reply":"2021-09-03T20:25:55.547456Z"},"trusted":true},"execution_count":199,"outputs":[{"name":"stdout","text":"Training set has 60816 samples and 369 features.\nTesting set has 15204 samples and 369 features.\nTrain target 0.03956%\nTest target 0.03959%\n","output_type":"stream"},{"execution_count":199,"output_type":"execute_result","data":{"text/plain":"104"},"metadata":{}}]},{"cell_type":"code","source":"#!pip install imblearn\nfrom imblearn.over_sampling import RandomOverSampler\nfrom imblearn.under_sampling import RandomUnderSampler\nrus = RandomUnderSampler(random_state=0)\nX_train = X_train[selected_features]\ny_train = y_train\nX_resampled, y_resampled = rus.fit_resample(X_train, y_train) \nfig, ax=plt.subplots(figsize=(8,6))\nsns.countplot(y_resampled);","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:25:58.943038Z","iopub.execute_input":"2021-09-03T20:25:58.943455Z","iopub.status.idle":"2021-09-03T20:25:59.188872Z","shell.execute_reply.started":"2021-09-03T20:25:58.943423Z","shell.execute_reply":"2021-09-03T20:25:59.187784Z"},"trusted":true},"execution_count":200,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 576x432 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"ros = RandomOverSampler(random_state = 0)\nX_rosampled, y_rosampled = ros.fit_resample(X_train,y_train) \nfig, ax = plt.subplots(figsize=(8,6))\nsns.countplot(y_rosampled);","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:26:02.044159Z","iopub.execute_input":"2021-09-03T20:26:02.044583Z","iopub.status.idle":"2021-09-03T20:26:02.934978Z","shell.execute_reply.started":"2021-09-03T20:26:02.044548Z","shell.execute_reply":"2021-09-03T20:26:02.933960Z"},"trusted":true},"execution_count":201,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 576x432 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"%%time\n\nX_train, X_test, y_train, y_test = train_test_split(X_resampled,y_resampled , test_size=0.20, random_state=1234)\nprint(X_test.shape, y_test.shape)\nprint(X_train.shape, y_train.shape)\n\n\n\nparam_grid = {'max_depth':[14, 16, 18],\n              'n_estimators':[40, 45, 47],\n              'criterion': ['gini'],\n              'min_samples_split':[int(X_train.shape[0]*0.01),\n                                   int(X_train.shape[0]*0.005)],\n              'max_features':['auto','log2',40]\n             }\n\n\nskf = StratifiedKFold(n_splits=5, shuffle = True, random_state = 1234)\n\ngsearch = GridSearchCV(RandomForestClassifier(random_state=1234, n_jobs=-1), \n                       param_grid={'max_depth':[14, 16, 18],\n                                   'n_estimators':[40, 45, 47],\n                                   'criterion': ['gini'],\n                                   'min_samples_split':[int(X_train.shape[0]*0.01),\n                                                       int(X_train.shape[0]*0.005)],\n                                   'max_features':['auto','log2',40]\n                                   }, \n                       scoring=max_profit_scorer, \n                       cv=skf.split(X_train, y_train), \n                       n_jobs=-1, \n                       verbose=5)\n\ngsearch.fit(X_train, y_train)\n\n\nbst_max_profit_model = gsearch.best_estimator_\nprint('treino:')\nshow_performance(bst_max_profit_model, X_train, y_train)\n\nprint('test:')\nshow_performance(bst_max_profit_model, X_test, y_test)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:27:47.828715Z","iopub.execute_input":"2021-09-03T20:27:47.829118Z","iopub.status.idle":"2021-09-03T20:28:32.413391Z","shell.execute_reply.started":"2021-09-03T20:27:47.829085Z","shell.execute_reply":"2021-09-03T20:28:32.412276Z"},"trusted":true},"execution_count":203,"outputs":[{"name":"stdout","text":"(963, 104) (963,)\n(3849, 104) (3849,)\nFitting 5 folds for each of 54 candidates, totalling 270 fits\n","output_type":"stream"},{"name":"stderr","text":"[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.\n[Parallel(n_jobs=-1)]: Done  10 tasks      | elapsed:    3.5s\n[Parallel(n_jobs=-1)]: Done  64 tasks      | elapsed:   10.2s\n[Parallel(n_jobs=-1)]: Done 154 tasks      | elapsed:   24.3s\n[Parallel(n_jobs=-1)]: Done 270 out of 270 | elapsed:   43.8s finished\n","output_type":"stream"},{"name":"stdout","text":"treino:\nMatriz de confusão:\n[[1605  296]\n [ 293 1655]]\n\nLucro obtido com a ação (em Reais): R$ 145990 \nLucro maximo possivel: R$ 175320 \nPorcentagem do lucro maximo: 83.27%\n\ntest:\nMatriz de confusão:\n[[388 117]\n [103 355]]\n\nLucro obtido com a ação (em Reais): R$ 30780 \nLucro maximo possivel: R$ 41220 \nPorcentagem do lucro maximo: 74.67%\n\nCPU times: user 2.86 s, sys: 196 ms, total: 3.05 s\nWall time: 44.6 s\n","output_type":"stream"}]},{"cell_type":"code","source":"%%time\n\nX_train, X_test, y_train, y_test = train_test_split(X_rosampled,y_rosampled , test_size=0.20, random_state=1234)\nprint(X_test.shape, y_test.shape)\nprint(X_train.shape, y_train.shape)\n\n\n\nskf = StratifiedKFold(n_splits=5, shuffle = True, random_state = 1234)\n\ngsearch = GridSearchCV(RandomForestClassifier(random_state=1234, n_jobs=-1), \n                       param_grid={'max_depth':[14, 16, 18],\n                                   'n_estimators':[40, 45, 47],\n                                   'criterion': ['gini'],\n                                   'min_samples_split':[int(X_train.shape[0]*0.01),\n                                                       int(X_train.shape[0]*0.005)],\n                                   'max_features':['auto','log2',40]\n                                   }, \n                       scoring=max_profit_scorer, \n                       cv=skf.split(X_train, y_train), \n                       n_jobs=-1, \n                       verbose=5)\n\ngsearch.fit(X_train, y_train)\n\n\nbst_max_profit_model = gsearch.best_estimator_\nprint('treino:')\nshow_performance(bst_max_profit_model, X_train, y_train)\n\nprint('test:')\nshow_performance(bst_max_profit_model, X_test, y_test)","metadata":{"execution":{"iopub.status.busy":"2021-09-03T20:28:44.803770Z","iopub.execute_input":"2021-09-03T20:28:44.804145Z"},"trusted":true},"execution_count":null,"outputs":[{"name":"stdout","text":"(23364, 104) (23364,)\n(93456, 104) (93456,)\nFitting 5 folds for each of 54 candidates, totalling 270 fits\n","output_type":"stream"},{"name":"stderr","text":"[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.\n[Parallel(n_jobs=-1)]: Done  10 tasks      | elapsed:   23.9s\n[Parallel(n_jobs=-1)]: Done  64 tasks      | elapsed:  2.2min\n","output_type":"stream"}]},{"cell_type":"markdown","source":"o resultado do nosso GridSearchCV nos da os melhores hiperparametros possiveis para o nosso melhor modelo, agora tomaremos alguma ações de feature enginering mantendo estes hiperparametros constantes a fim de determinar se nossas ações melhoraram/atrapalharam nosso modelo.","metadata":{}},{"cell_type":"markdown","source":"- A primeira ação será dar uma olhada rapida nas categorias possiveis sobre cada coluna, isso nos permitira identificar variaveis sem variancia, e também teremos uma leve noção de quais variaveis podem ser categoricas e quais numericas.","metadata":{}},{"cell_type":"markdown","source":"- de fato existem features sem variancia, estas não aportam informação discrimante para o nosso modelo, pelo qual procederemos a retira-las do dataset","metadata":{}},{"cell_type":"markdown","source":"- A segunda ação que executaremos será de remover variaveis duplicadas. Para isso barremos cada feature comparandola com todas as outras, esta comparação será feita pela utilização da função array_equal do numpy\n- Comparamos pares de variaveis em ordem ascendente, se é comprovado que existem variaveis duplicadas adicinaremos a segunda a um diccionario, aqui tomamos o cuidado de não executar a comparação sobre variaveis que já foram identificadas para ser removidas (evitemos dores de cabeça).","metadata":{}},{"cell_type":"markdown","source":"mmm, aqui é um bom momento para refletir, nosso modelo piorou mas ele possui menos variaveis, só foram retiradas variaveis duplicadas tomando o cuidado de deixar pelo menos 1 das que possuim a informação duplicada. Vale a pena deixar variaveis duplicadas sendo que isso não faz sentido? acho que não... Continuareremos com o modelo só com 1 das variaveis duplicadas. Por que? porque desde o ponto de vista conceitual, não existe motivo algum para ter um modelo melhor só pelo fato de ter replicado algumas colunas. OK? Não ficou claro? acredito fortemente que mexendo nos hiperparametros este dataset dará igual ou melhor resultado que o anterior que possui variaveis duplicadas. Vamos na frente.","metadata":{}},{"cell_type":"markdown","source":"OK, OK, pensemos no que pode estar acontecendo. Antes de entrar no desespero pensemos, estamos diminuindo a variancia do nosso modelo, porque está melhorando no treindo e piorando no teste? pode ser que estamos tirando informação importante, mas como pode ser que estamos tirando informação se a variancia é quase zero. Ok não será que estamos esquecendo de alguma coisa... tipo: outliers? testemos tirar eles do nosso dataset","metadata":{}},{"cell_type":"markdown","source":"- Apos retirar as variaveis duplicadas vemos que aumentou um pouco mais o porcentagem do lucro ideal","metadata":{}},{"cell_type":"markdown","source":"- identificamos algumas variaveis com variancia 0, retiramos estas variaveis do dataset dado que não aportam nenhum tipo de informação\n\n- identificamos algumas variaveis binarias, outras categoricas e demais continuas, separamos elas para respectivas análises","metadata":{}},{"cell_type":"markdown","source":"- quais variáveis serão dummizadas -> exclusão das que possuem muitas categorias","metadata":{}},{"cell_type":"markdown","source":"- Método para limpar colunas da base treino identificando as variáveis categóricas com variância próxima de zero, no caso todas em que a relação entre a primeira e segunda categoria é de 99 para 1.","metadata":{}},{"cell_type":"markdown","source":"# Referências\n- Trabalho do Caio Martins https://github.com/CaioMar/certificao_data_masters/blob/master/Certificacao%20Data%20Masters.ipynb\n- Trabalho do PedroHCouto https://github.com/PedroHCouto/Santander-Case/blob/master/Part%20A%20-%20Classification.ipynb\n- Random Forest Classifier and its Hyperparameters https://medium.com/analytics-vidhya/random-forest-classifier-and-its-hyperparameters-8467bec755f6\n- Hyperparameter Tuning the Random Forest in Python https://towardsdatascience.com/hyperparameter-tuning-the-random-forest-in-python-using-scikit-learn-28d2aa77dd74\n- Random Search vs Grid Search for hyperparameter optimization https://towardsdatascience.com/random-search-vs-grid-search-for-hyperparameter-optimization-345e1422899d\n- Optimizing Hyperparameters the right Way https://towardsdatascience.com/optimizing-hyperparameters-the-right-way-3c9cafc279cc\n- How to Use ROC Curves and Precision-Recall Curves for Classification in Python\nhttps://machinelearningmastery.com/roc-curves-and-precision-recall-curves-for-classification-in-python/\n- Step by Step to Understanding K-means Clustering and Implementation with sklearn https://medium.com/data-folks-indonesia/step-by-step-to-understanding-k-means-clustering-and-implementation-with-sklearn-b55803f519d6","metadata":{}}]}