notebook/ML_breast_cancer_prediction/knn-breast-cancer-prediction.ipynb

{"cells": [{"metadata": {}, "cell_type": "markdown", "source": "# 0.\u6570\u636e\u96c6\u4ecb\u7ecd\u4e0e\u5b9e\u9a8c\u51c6\u5907\n\n## \u6570\u636e\u96c6\u4ecb\u7ecd\n\u8be5\u6570\u636e\u662f\u5a01\u65af\u5eb7\u661f\u5dde\u4e73\u817a\u764c\u8bca\u65ad\u6570\u636e\u5e93\uff0c\u91cc\u9762\u5171\u6709699\u4e2a\u6837\u672c\uff0c\u6bcf\u4e2a\u6837\u672c\u670910\u4e2a\u7279\u5f81\uff08\u5176\u4e2d1\u4e2a\u662f\u7edf\u8ba1\u7f16\u53f7\uff09\uff0c1\u4e2a\u6807\u7b7e\u3002\u6837\u672c\u7279\u5f81\u542b\u4e49\u53ca\u53d6\u503c\u8303\u56f4\u5982\u4e0b\uff1a\n1.\u7f16\u53f7\n2.\u5757\u539a\u5ea6\n3.\u7ec6\u80de\u5927\u5c0f\u7684\u4e00\u81f4\u6027\n4.\u7ec6\u80de\u5f62\u72b6\u7684\u5747\u5300\u6027\n5.\u8fb9\u7f18\u9644\u7740\u529b\n6.\u5355\u4e2a\u4e0a\u76ae\u7ec6\u80de\u5927\u5c0f\n7.\u88f8\u6838\n8.\u67d3\u8272\u8d28\n9.\u6b63\u5e38\u6838\u4ec1\n10.\u6709\u4e1d\u5206\u88c2\n11.\u5206\u7c7b\uff1a\uff082\u4e3a\u826f\u6027\uff0c4\u4e3a\u6076\u6027\uff09\n\u7279\u5f812-10\u7684\u53d6\u503c\u8303\u56f4\u5747\u4e3a[1,10]"}, {"metadata": {}, "cell_type": "markdown", "source": "## \u5b9e\u9a8c\u51c6\u5907\n\n### \u8fdb\u5165ModelArts\u754c\u9762\n\n\u70b9\u51fb\u5982\u4e0b\u94fe\u63a5\uff1ahttps://www.huaweicloud.com/product/modelarts.html \uff0c\u8fdb\u5165ModelArts\u4e3b\u9875\u3002\u70b9\u51fb\u201c\u7acb\u5373\u4f7f\u7528\u201d\u6309\u94ae\uff0c\u8f93\u5165\u7528\u6237\u540d\u548c\u5bc6\u7801\u767b\u5f55\uff0c\u8fdb\u5165ModelArts\u4f7f\u7528\u9875\u9762\u3002\n\n### \u521b\u5efaModelArts Notebook\n\n\u4e0b\u9762\uff0c\u6211\u4eec\u5728ModelArts\u4e2d\u521b\u5efa\u4e00\u4e2anotebook\u5f00\u53d1\u73af\u5883\uff0cModelArts notebook\u63d0\u4f9b\u7f51\u9875\u7248\u7684Python\u5f00\u53d1\u73af\u5883\uff0c\u53ef\u4ee5\u65b9\u4fbf\u7684\u7f16\u5199\u3001\u8fd0\u884c\u4ee3\u7801\uff0c\u5e76\u67e5\u770b\u8fd0\u884c\u7ed3\u679c\u3002\n\n\u7b2c\u4e00\u6b65\uff1a\u5728ModelArts\u670d\u52a1\u4e3b\u754c\u9762\u4f9d\u6b21\u70b9\u51fb\u201c\u5f00\u53d1\u73af\u5883\u201d\u3001\u201c\u521b\u5efa\u201d\n\n![create_nb_create_button](./img/create_nb_create_button.png)\n\n\u7b2c\u4e8c\u6b65\uff1a\u586b\u5199notebook\u6240\u9700\u7684\u53c2\u6570\uff1a\n\n| \u53c2\u6570 | \u8bf4\u660e |\n| - - - - - | - - - - - |\n| \u8ba1\u8d39\u65b9\u5f0f | \u6309\u9700\u8ba1\u8d39  |\n| \u540d\u79f0 | Notebook\u5b9e\u4f8b\u540d\u79f0\uff0c\u5982 kmeans_customer_segmentation |\n| \u5de5\u4f5c\u73af\u5883 | Python3 |\n| \u8d44\u6e90\u6c60 | \u9009\u62e9\"\u516c\u5171\u8d44\u6e90\u6c60\"\u5373\u53ef |\n| \u7c7b\u578b | CPU |\n| \u89c4\u683c | 2\u68388GiB |\n| \u5b58\u50a8\u914d\u7f6e | \u9009\u62e9EVS\uff0c\u78c1\u76d8\u89c4\u683c5GB |\n\n\u7b2c\u4e09\u6b65\uff1a\u914d\u7f6e\u597dnotebook\u53c2\u6570\u540e\uff0c\u70b9\u51fb\u4e0b\u4e00\u6b65\uff0c\u8fdb\u5165notebook\u4fe1\u606f\u9884\u89c8\u3002\u786e\u8ba4\u65e0\u8bef\u540e\uff0c\u70b9\u51fb\u201c\u7acb\u5373\u521b\u5efa\u201d\n\n![create_nb_creation_summary](./img/create_nb_creation_summary.png)\n\n\u7b2c\u56db\u6b65\uff1a\u521b\u5efa\u5b8c\u6210\u540e\uff0c\u8fd4\u56de\u5f00\u53d1\u73af\u5883\u4e3b\u754c\u9762\uff0c\u7b49\u5f85Notebook\u521b\u5efa\u5b8c\u6bd5\u540e\uff0c\u6253\u5f00Notebook\uff0c\u8fdb\u884c\u4e0b\u4e00\u6b65\u64cd\u4f5c\u3002\n![modelarts_notebook_index](./img/modelarts_notebook_index.png)\n\n### \u5728ModelArts\u4e2d\u521b\u5efa\u5f00\u53d1\u73af\u5883\n\n\u63a5\u4e0b\u6765\uff0c\u6211\u4eec\u521b\u5efa\u4e00\u4e2a\u5b9e\u9645\u7684\u5f00\u53d1\u73af\u5883\uff0c\u7528\u4e8e\u540e\u7eed\u7684\u5b9e\u9a8c\u6b65\u9aa4\u3002\n\n\u7b2c\u4e00\u6b65\uff1a\u70b9\u51fb\u4e0b\u56fe\u6240\u793a\u7684\u201c\u6253\u5f00\u201d\u6309\u94ae\uff0c\u8fdb\u5165\u521a\u521a\u521b\u5efa\u7684Notebook\n![inter_dev_env](img/enter_dev_env.png)\n\n\u7b2c\u4e8c\u6b65\uff1a\u521b\u5efa\u4e00\u4e2aPython3\u73af\u5883\u7684\u7684Notebook\u3002\u70b9\u51fb\u53f3\u4e0a\u89d2\u7684\"New\"\uff0c\u7136\u540e\u521b\u5efa XGBoost-Sklearn \u5f00\u53d1\u73af\u5883\u3002\n\n\u7b2c\u4e09\u6b65\uff1a\u70b9\u51fb\u5de6\u4e0a\u65b9\u7684\u6587\u4ef6\u540d\"Untitled\"\uff0c\u5e76\u8f93\u5165\u4e00\u4e2a\u4e0e\u672c\u5b9e\u9a8c\u76f8\u5173\u7684\u540d\u79f0\uff0c\u5982\"knn-breast-cancer-classfication\"\n![notebook_untitled_filename](./img/notebook_untitled_filename.png)\n![notebook_name_the_ipynb](./img/notebook_name_the_ipynb.png)\n\n\n### \u5728Notebook\u4e2d\u7f16\u5199\u5e76\u6267\u884c\u4ee3\u7801\n\n\u5728Notebook\u4e2d\uff0c\u6211\u4eec\u8f93\u5165\u4e00\u4e2a\u7b80\u5355\u7684\u6253\u5370\u8bed\u53e5\uff0c\u7136\u540e\u70b9\u51fb\u4e0a\u65b9\u7684\u8fd0\u884c\u6309\u94ae\uff0c\u53ef\u4ee5\u67e5\u770b\u8bed\u53e5\u6267\u884c\u7684\u7ed3\u679c\uff1a\n![run_helloworld](./img/run_helloworld.png)\n\n\n\u5f00\u53d1\u73af\u5883\u51c6\u5907\u597d\u5566\uff0c\u63a5\u4e0b\u6765\u53ef\u4ee5\u6109\u5feb\u5730\u5199\u4ee3\u7801\u5566\uff01\n"}, {"metadata": {}, "cell_type": "markdown", "source": "## \u5bfc\u5165\u7c7b\u5e93"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "import pandas as pd\nfrom pandas import read_csv\nfrom pandas.plotting import scatter_matrix\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport seaborn as sns \nimport requests \nfrom sklearn import datasets\nfrom matplotlib import pyplot\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.model_selection import KFold\nfrom sklearn.model_selection import cross_val_score\nfrom sklearn.metrics import classification_report\nfrom sklearn.metrics import confusion_matrix\nfrom sklearn.metrics import accuracy_score\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.tree import DecisionTreeClassifier\nfrom sklearn.discriminant_analysis import LinearDiscriminantAnalysis\nfrom sklearn.neighbors import KNeighborsClassifier\nfrom sklearn.naive_bayes import GaussianNB\nfrom sklearn.svm import SVC\nfrom sklearn.preprocessing import LabelEncoder\nfrom sklearn.linear_model import LogisticRegression", "execution_count": 256, "outputs": []}, {"metadata": {}, "cell_type": "markdown", "source": "# 1.\u6570\u636e\u9884\u5904\u7406"}, {"metadata": {}, "cell_type": "markdown", "source": "## 1.1\u4e0b\u8f7d\u6570\u636e\u5e76\u663e\u793a\u524d5\u884c\n\u672c\u6837\u4f8b\u6570\u636e\u5df2\u7ecf\u9884\u5148\u88ab\u4e0a\u4f20\u5230\u516c\u5171OBS\u6876\u91cc\uff0c\u6211\u4eec\u53ea\u9700\u8981\u4e0b\u8f7d\u8bfb\u53d6\u5373\u53ef"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "url = 'https://modelarts-labs.obs.cn-north-1.myhuaweicloud.com/notebook/ML_breast_cancer_wisconsin/breast-cancer-wisconsin.data'\nr = requests.get(url) \nwith open(\"./breast-cancer-wisconsin.data\", \"wb\") as code:\n    code.write(r.content)\nbreast_cancer_data =pd.read_csv('./breast-cancer-wisconsin.data',header=None\n                               ,names = ['numbers','Clump_Thickness','Uniformity_of_Cell_Size','Uniformity_of_Cell_Shape','Marginal_Adhesion',\n                                         'Single_Epithelial_Cell_Size'\n                                        ,'Bare_Nuclei','Bland_Chromatin','Normal_Nucleoli','Mitoses','Class'])\nbreast_cancer_data.head(5)", "execution_count": 255, "outputs": [{"output_type": "execute_result", "execution_count": 255, "data": {"text/plain": "   numbers  Clump_Thickness  Uniformity_of_Cell_Size  \\\n0  1000025                5                        1   \n1  1002945                5                        4   \n2  1015425                3                        1   \n3  1016277                6                        8   \n4  1017023                4                        1   \n\n   Uniformity_of_Cell_Shape  Marginal_Adhesion  Single_Epithelial_Cell_Size  \\\n0                         1                  1                            2   \n1                         4                  5                            7   \n2                         1                  1                            2   \n3                         8                  1                            3   \n4                         1                  3                            2   \n\n  Bare_Nuclei  Bland_Chromatin  Normal_Nucleoli  Mitoses  Class  \n0           1                3                1        1      2  \n1          10                3                2        1      2  \n2           2                3                1        1      2  \n3           4                3                7        1      2  \n4           1                3                1        1      2  ", "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>numbers</th>\n      <th>Clump_Thickness</th>\n      <th>Uniformity_of_Cell_Size</th>\n      <th>Uniformity_of_Cell_Shape</th>\n      <th>Marginal_Adhesion</th>\n      <th>Single_Epithelial_Cell_Size</th>\n      <th>Bare_Nuclei</th>\n      <th>Bland_Chromatin</th>\n      <th>Normal_Nucleoli</th>\n      <th>Mitoses</th>\n      <th>Class</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1000025</td>\n      <td>5</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>2</td>\n      <td>1</td>\n      <td>3</td>\n      <td>1</td>\n      <td>1</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>1002945</td>\n      <td>5</td>\n      <td>4</td>\n      <td>4</td>\n      <td>5</td>\n      <td>7</td>\n      <td>10</td>\n      <td>3</td>\n      <td>2</td>\n      <td>1</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>1015425</td>\n      <td>3</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>2</td>\n      <td>2</td>\n      <td>3</td>\n      <td>1</td>\n      <td>1</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>1016277</td>\n      <td>6</td>\n      <td>8</td>\n      <td>8</td>\n      <td>1</td>\n      <td>3</td>\n      <td>4</td>\n      <td>3</td>\n      <td>7</td>\n      <td>1</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>1017023</td>\n      <td>4</td>\n      <td>1</td>\n      <td>1</td>\n      <td>3</td>\n      <td>2</td>\n      <td>1</td>\n      <td>3</td>\n      <td>1</td>\n      <td>1</td>\n      <td>2</td>\n    </tr>\n  </tbody>\n</table>\n</div>"}, "metadata": {}}]}, {"metadata": {"scrolled": true, "trusted": true}, "cell_type": "code", "source": "data=breast_cancer_data\ndata.head()\ndata=data.drop(['numbers'], axis=1)  # \u5c06\u5e8f\u53f7\u5217\u53bb\u6389\nlen(data)", "execution_count": 251, "outputs": [{"output_type": "execute_result", "execution_count": 251, "data": {"text/plain": "699"}, "metadata": {}}]}, {"metadata": {}, "cell_type": "markdown", "source": "## 1.2\u6570\u503c\u5c5e\u6027\u6982\u62ec\n\ndescribe()\u65b9\u6cd5\u53ef\u4ee5\u5c55\u793a\u6570\u503c\u5c5e\u6027\u7684\u6982\u62ec\u3002\n\u5176\u4e2d,count\u662f\u975e\u7a7a\u503c\u7684\u603b\u6570\u91cf\uff0cmean\u662f\u5e73\u5747\u6570\uff0cstd\u662f\u6807\u51c6\u5dee\uff0c25%\uff0c50%\uff0c75%\u8868\u793a\u5bf9\u5440\u7684\u5206\u4f4d\u6570\u3002"}, {"metadata": {"scrolled": false, "trusted": true}, "cell_type": "code", "source": "data.describe()", "execution_count": 124, "outputs": [{"output_type": "execute_result", "execution_count": 124, "data": {"text/plain": "       Clump_Thickness  Uniformity_of_Cell_Size  Uniformity_of_Cell_Shape  \\\ncount       699.000000               699.000000                699.000000   \nmean          4.417740                 3.134478                  3.207439   \nstd           2.815741                 3.051459                  2.971913   \nmin           1.000000                 1.000000                  1.000000   \n25%           2.000000                 1.000000                  1.000000   \n50%           4.000000                 1.000000                  1.000000   \n75%           6.000000                 5.000000                  5.000000   \nmax          10.000000                10.000000                 10.000000   \n\n       Marginal_Adhesion  Single_Epithelial_Cell_Size  Bland_Chromatin  \\\ncount         699.000000                   699.000000       699.000000   \nmean            2.806867                     3.216023         3.437768   \nstd             2.855379                     2.214300         2.438364   \nmin             1.000000                     1.000000         1.000000   \n25%             1.000000                     2.000000         2.000000   \n50%             1.000000                     2.000000         3.000000   \n75%             4.000000                     4.000000         5.000000   \nmax            10.000000                    10.000000        10.000000   \n\n       Normal_Nucleoli     Mitoses       Class  \ncount       699.000000  699.000000  699.000000  \nmean          2.866953    1.589413    2.689557  \nstd           3.053634    1.715078    0.951273  \nmin           1.000000    1.000000    2.000000  \n25%           1.000000    1.000000    2.000000  \n50%           1.000000    1.000000    2.000000  \n75%           4.000000    1.000000    4.000000  \nmax          10.000000   10.000000    4.000000  ", "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>Clump_Thickness</th>\n      <th>Uniformity_of_Cell_Size</th>\n      <th>Uniformity_of_Cell_Shape</th>\n      <th>Marginal_Adhesion</th>\n      <th>Single_Epithelial_Cell_Size</th>\n      <th>Bland_Chromatin</th>\n      <th>Normal_Nucleoli</th>\n      <th>Mitoses</th>\n      <th>Class</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>count</th>\n      <td>699.000000</td>\n      <td>699.000000</td>\n      <td>699.000000</td>\n      <td>699.000000</td>\n      <td>699.000000</td>\n      <td>699.000000</td>\n      <td>699.000000</td>\n      <td>699.000000</td>\n      <td>699.000000</td>\n    </tr>\n    <tr>\n      <th>mean</th>\n      <td>4.417740</td>\n      <td>3.134478</td>\n      <td>3.207439</td>\n      <td>2.806867</td>\n      <td>3.216023</td>\n      <td>3.437768</td>\n      <td>2.866953</td>\n      <td>1.589413</td>\n      <td>2.689557</td>\n    </tr>\n    <tr>\n      <th>std</th>\n      <td>2.815741</td>\n      <td>3.051459</td>\n      <td>2.971913</td>\n      <td>2.855379</td>\n      <td>2.214300</td>\n      <td>2.438364</td>\n      <td>3.053634</td>\n      <td>1.715078</td>\n      <td>0.951273</td>\n    </tr>\n    <tr>\n      <th>min</th>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>2.000000</td>\n    </tr>\n    <tr>\n      <th>25%</th>\n      <td>2.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>2.000000</td>\n      <td>2.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>2.000000</td>\n    </tr>\n    <tr>\n      <th>50%</th>\n      <td>4.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>2.000000</td>\n      <td>3.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>2.000000</td>\n    </tr>\n    <tr>\n      <th>75%</th>\n      <td>6.000000</td>\n      <td>5.000000</td>\n      <td>5.000000</td>\n      <td>4.000000</td>\n      <td>4.000000</td>\n      <td>5.000000</td>\n      <td>4.000000</td>\n      <td>1.000000</td>\n      <td>4.000000</td>\n    </tr>\n    <tr>\n      <th>max</th>\n      <td>10.000000</td>\n      <td>10.000000</td>\n      <td>10.000000</td>\n      <td>10.000000</td>\n      <td>10.000000</td>\n      <td>10.000000</td>\n      <td>10.000000</td>\n      <td>10.000000</td>\n      <td>4.000000</td>\n    </tr>\n  </tbody>\n</table>\n</div>"}, "metadata": {}}]}, {"metadata": {}, "cell_type": "markdown", "source": "## 1.3\u6570\u636e\u7684\u63cf\u8ff0\uff08\u603b\u884c\u6570\u3001\u6bcf\u4e2a\u5c5e\u6027\u7684\u7c7b\u578b\u3001\u975e\u7a7a\u503c\u6570\u91cf\uff09\ninfo()\u65b9\u6cd5\u53ef\u4ee5\u5feb\u901f\u67e5\u770b\u6570\u636e\u7684\u63cf\u8ff0\uff0c\u7279\u522b\u662f\u603b\u884c\u6570\u3001\u6bcf\u4e2a\u5c5e\u6027\u7684\u7c7b\u578b\u548c\u975e\u7a7a\u503c\u7684\u6570\u91cf\u3002"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "data.info()", "execution_count": 125, "outputs": [{"output_type": "stream", "text": "<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 699 entries, 0 to 698\nData columns (total 10 columns):\nClump_Thickness                699 non-null int64\nUniformity_of_Cell_Size        699 non-null int64\nUniformity_of_Cell_Shape       699 non-null int64\nMarginal_Adhesion              699 non-null int64\nSingle_Epithelial_Cell_Size    699 non-null int64\nBare_Nuclei                    699 non-null object\nBland_Chromatin                699 non-null int64\nNormal_Nucleoli                699 non-null int64\nMitoses                        699 non-null int64\nClass                          699 non-null int64\ndtypes: int64(9), object(1)\nmemory usage: 54.7+ KB\n", "name": "stdout"}]}, {"metadata": {}, "cell_type": "markdown", "source": "## 1.3\u5f02\u5e38\u503c\u5904\u7406"}, {"metadata": {}, "cell_type": "markdown", "source": "\u6211\u4eec\u53ef\u4ee5\u4ece\u4e0a\u9762\u770b\u5230\uff0c\u6570\u636e\u91cc\u201cBare_Nuclei\u201d\u8fd9\u4e00\u5217\u662fobject\u7279\u5f81\uff0c\u6211\u4eec\u53ef\u4ee5\u6253\u5370\u4e00\u4e0b\u8fd9\u4e00\u5217\u89c2\u5bdf\u4e00\u4e0b\u6570\u636e\uff1a"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "for index,row in data.iterrows():\n    print(row['Bare_Nuclei'])", "execution_count": 126, "outputs": [{"output_type": "stream", "text": "1\n10\n2\n4\n1\n10\n10\n1\n1\n1\n1\n1\n3\n3\n9\n1\n1\n1\n10\n1\n10\n7\n1\n?\n1\n7\n1\n1\n1\n1\n1\n1\n5\n1\n1\n1\n1\n1\n10\n7\n?\n3\n10\n1\n1\n1\n9\n1\n1\n8\n3\n4\n5\n8\n8\n5\n6\n1\n10\n2\n3\n2\n8\n2\n1\n2\n1\n10\n9\n1\n1\n2\n1\n10\n4\n2\n1\n1\n3\n1\n1\n1\n1\n2\n9\n4\n8\n10\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n6\n10\n5\n5\n1\n3\n1\n3\n10\n10\n1\n9\n2\n9\n10\n8\n3\n5\n2\n10\n3\n2\n1\n2\n10\n10\n7\n1\n10\n1\n10\n1\n1\n1\n10\n1\n1\n2\n1\n1\n1\n?\n1\n1\n5\n5\n1\n?\n8\n2\n1\n10\n1\n10\n5\n3\n1\n10\n1\n1\n?\n10\n10\n1\n1\n3\n?\n2\n10\n1\n1\n1\n1\n1\n1\n10\n10\n10\n1\n1\n1\n10\n1\n1\n1\n10\n10\n1\n8\n10\n8\n1\n8\n10\n1\n1\n1\n1\n7\n1\n1\n1\n10\n10\n1\n1\n1\n10\n5\n1\n1\n1\n10\n8\n1\n10\n10\n5\n1\n1\n4\n1\n1\n10\n5\n8\n10\n1\n10\n5\n1\n10\n7\n8\n1\n10\n1\n?\n10\n2\n9\n10\n2\n1\n1\n5\n1\n2\n10\n9\n1\n?\n1\n10\n10\n10\n8\n10\n1\n1\n1\n8\n10\n10\n10\n10\n3\n1\n10\n10\n4\n1\n10\n1\n10\n4\n1\n?\n1\n1\n1\n7\n1\n1\n10\n10\n10\n10\n10\n1\n5\n10\n1\n1\n?\n10\n?\n10\n5\n?\n1\n10\n4\n1\n10\n1\n10\n10\n1\n1\n3\n5\n1\n1\n1\n1\n1\n?\n10\n8\n1\n5\n10\n?\n1\n10\n1\n1\n10\n1\n4\n10\n8\n1\n1\n10\n10\n1\n10\n1\n1\n10\n10\n1\n1\n1\n10\n1\n1\n1\n1\n8\n1\n1\n3\n10\n1\n1\n3\n10\n4\n7\n10\n10\n3\n3\n1\n1\n10\n10\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n10\n1\n1\n1\n1\n10\n1\n1\n2\n1\n10\n1\n1\n1\n1\n1\n1\n1\n1\n9\n1\n1\n4\n1\n1\n1\n1\n2\n1\n1\n?\n4\n1\n10\n3\n10\n1\n2\n1\n3\n10\n1\n1\n1\n10\n1\n2\n1\n1\n1\n1\n1\n1\n8\n10\n1\n1\n1\n1\n10\n4\n3\n2\n1\n1\n1\n1\n1\n10\n1\n1\n1\n10\n1\n6\n10\n3\n1\n1\n1\n5\n1\n1\n1\n4\n10\n10\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n10\n1\n1\n5\n10\n1\n3\n1\n10\n3\n4\n1\n10\n1\n10\n5\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n5\n4\n1\n1\n1\n1\n1\n1\n10\n10\n1\n1\n1\n10\n1\n1\n5\n10\n1\n1\n1\n1\n1\n1\n10\n1\n1\n1\n1\n1\n1\n1\n1\n1\n2\n1\n1\n1\n1\n1\n10\n1\n1\n5\n1\n1\n1\n5\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n10\n1\n3\n10\n5\n10\n10\n1\n1\n2\n1\n1\n1\n1\n1\n1\n10\n10\n1\n1\n1\n10\n1\n3\n1\n1\n10\n10\n1\n10\n1\n1\n1\n1\n1\n1\n1\n1\n1\n10\n8\n1\n1\n10\n1\n10\n2\n10\n1\n1\n1\n1\n?\n1\n1\n1\n2\n1\n1\n1\n4\n6\n5\n1\n1\n1\n1\n1\n3\n1\n1\n1\n2\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n2\n1\n4\n1\n1\n1\n1\n1\n1\n1\n10\n1\n1\n1\n1\n1\n1\n1\n1\n1\n1\n5\n8\n1\n1\n1\n1\n1\n1\n1\n1\n1\n10\n10\n1\n1\n1\n1\n1\n1\n1\n1\n1\n5\n1\n1\n2\n1\n3\n4\n5\n", "name": "stdout"}]}, {"metadata": {}, "cell_type": "markdown", "source": "\u6211\u4eec\u53ef\u4ee5\u89c2\u5bdf\u5230\uff0c\u9664\u4e86\u6570\u5b57\u5916\uff0c\u8fd8\u6709\u4e00\u4e9b\u5b57\u7b26\u578b\u7684\uff0c\u6bd4\u5982\u95ee\u53f7\uff0c\u53ef\u80fd\u662f\u6536\u96c6\u6570\u636e\u7684\u65f6\u5019\u7edf\u8ba1\u4e0d\u5168\uff0c\u6211\u4eec\u73b0\u5728\u627e\u5230\u8fd9\u90e8\u5206\u7279\u5f81\uff0c\u6253\u5370\u51fa\u8fd9\u90e8\u5206\u5b57\u7b26\u548c\u5176\u4f4d\u7f6e\uff0c\u5e76\u7528\u5747\u503c\u66ff\u4ee3\u5b83\u4eec\u3002"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "Bare_Nuclei=[]\ncount_null=0\nfor index,row in data.iterrows():\n    try:\n        Bare_Nuclei.append(int(row['Bare_Nuclei']))\n    except:\n        count_null+=1\n        Bare_Nuclei.append(0)\n        print(row['Bare_Nuclei'],index)\nBare_Nuclei_np=np.array(Bare_Nuclei)\nBare_Nuclei_mean=np.sum(Bare_Nuclei)//(len(Bare_Nuclei)-count_null)\nfor i in range(len(Bare_Nuclei)):\n    if Bare_Nuclei[i]==0:\n        Bare_Nuclei[i]=Bare_Nuclei_mean\n    Bare_Nuclei[i]=int(Bare_Nuclei[i])\ndata['Bare_Nuclei']=Bare_Nuclei", "execution_count": 127, "outputs": [{"output_type": "stream", "text": "? 23\n? 40\n? 139\n? 145\n? 158\n? 164\n? 235\n? 249\n? 275\n? 292\n? 294\n? 297\n? 315\n? 321\n? 411\n? 617\n", "name": "stdout"}]}, {"metadata": {}, "cell_type": "markdown", "source": "\u518d\u6b21\u67e5\u770b\u6570\u636e\u63cf\u8ff0\uff0c\u5982\u4e0b\uff0c\u53ef\u4ee5\u770b\u5230\uff0c\u6bcf\u4e00\u5217\u90fd\u53d8\u6210\u6570\u503c\u7279\u5f81\u4e86"}, {"metadata": {"scrolled": true, "trusted": true}, "cell_type": "code", "source": "data.info()", "execution_count": 128, "outputs": [{"output_type": "stream", "text": "<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 699 entries, 0 to 698\nData columns (total 10 columns):\nClump_Thickness                699 non-null int64\nUniformity_of_Cell_Size        699 non-null int64\nUniformity_of_Cell_Shape       699 non-null int64\nMarginal_Adhesion              699 non-null int64\nSingle_Epithelial_Cell_Size    699 non-null int64\nBare_Nuclei                    699 non-null int64\nBland_Chromatin                699 non-null int64\nNormal_Nucleoli                699 non-null int64\nMitoses                        699 non-null int64\nClass                          699 non-null int64\ndtypes: int64(10)\nmemory usage: 54.7 KB\n", "name": "stdout"}]}, {"metadata": {}, "cell_type": "markdown", "source": "# 2.\u6570\u636e\u76f8\u5173\u6027\u5206\u6790"}, {"metadata": {}, "cell_type": "markdown", "source": "## 2.1\u6570\u636e\u7c7b\u522b\u7edf\u8ba1\n\u7c7b\u522b\u4e3a2\uff08\u5373\u672a\u60a3\u764c\u75c7\uff09\u7684\u70ed\u4e0d\u662f\u6545\u610f\u548c\u7c7b\u522b\u4e3a4\uff08\u5373\u60a3\u764c\u75c7\uff09\u7684\u4eba\u6570\u7edf\u8ba1\u548c\u793a\u610f\u56fe\u5982\u4e0b\uff1a"}, {"metadata": {"scrolled": true, "trusted": true}, "cell_type": "code", "source": "print(breast_cancer_data.groupby('Class').size())\nplt.figure(1 , figsize = (10 , 3))\nsns.countplot(y = 'Class' , data = data)\nplt.show()", "execution_count": 129, "outputs": [{"output_type": "stream", "text": "Class\n2    458\n4    241\ndtype: int64\n", "name": "stdout"}, {"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f1456f908>", "image/png": "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\n"}, "metadata": {}}]}, {"metadata": {}, "cell_type": "markdown", "source": "## 2.2\u5404\u7c7b\u522b\u7279\u5f81\u60c5\u51b5\u7edf\u8ba1\n\u4e0b\u56fe\u8868\u793a\u60a3\u764c\u8005\u548c\u672a\u60a3\u764c\u8005\u6bcf\u4e00\u7c7b\u7279\u5f81\u7684\u5e73\u5747\u503c\uff0c\u53ef\u4ee5\u770b\u5230\uff0c\u60a3\u764c\u8005\u7684\u6bcf\u4e00\u7c7b\u7279\u5f81\u90fd\u666e\u904d\u6bd4\u8f83\u9ad8\uff0c\u672a\u60a3\u764c\u8005\u90fd\u666e\u904d\u6bd4\u8f83\u4f4e\uff0c\u76f8\u5173\u6027\u8f83\u5f3a"}, {"metadata": {"scrolled": true, "trusted": true}, "cell_type": "code", "source": "fig,axes = plt.subplots(3,3,figsize=(20,15))\ndraw_column=['Clump_Thickness','Uniformity_of_Cell_Size','Uniformity_of_Cell_Shape','Marginal_Adhesion',\n                                         'Single_Epithelial_Cell_Size'\n                                        ,'Bare_Nuclei','Bland_Chromatin','Normal_Nucleoli','Mitoses']\nax_list=[axes[0,0],axes[0,1],axes[0,2],axes[1,0],axes[1,1],axes[1,2],axes[2,0],axes[2,1],axes[2,2]]\nfor i,column in enumerate(draw_column):\n    sns.barplot(y=column,x='Class',data=data,ax=ax_list[i])\nplt.show()\nplt.close()\n#     sns.barplot(x='Uniformity of Cell Size',y='Class',data=data,ax=axes[)\n#     sns.barplot(x='Marginal Adhesion',y='Class',data=data,ax=axes[0,2])\n# sns.barplot(x='Marginal Adhesion',y='Uniformity of Cell Size',hue='Class',data=data,ax=axes[1,1])\n# sns.barplot(y='Bare Nuclei',x='Class',data=data,ax=axes[2,1])", "execution_count": 130, "outputs": [{"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f14629f98>", "image/png": "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\n"}, "metadata": {}}]}, {"metadata": {}, "cell_type": "markdown", "source": "##  2.3\u7cfb\u6570\u77e9\u9635\u7ed8\u5236 \n\u76f8\u5173\u7cfb\u6570\u77e9\u9635\uff0c\u5e38\u7528\u7684\u6709\u76ae\u5c14\u900a\u76f8\u5173\u7cfb\u6570(Pearson product-moment correlation coefficient\uff0cPearson's r)\u7684\u65b9\u9635\uff0c\u901a\u8fc7\u5b83\u53ef\u4ee5\u6765\u8861\u91cf\u4e24\u4e2a\u7279\u5f81\u4e4b\u95f4\u7684\u7ebf\u6027\u5173\u7cfb\u3002\u76ae\u5c14\u900a\u76f8\u5173\u7cfb\u6570\u7684\u53d6\u503c\u5728[-1,1]\u8303\u56f4\u5185\uff0c\u5982\u679cr=1\uff0c\u8868\u793a\u4e24\u4e2a\u53d8\u91cf\u5448\u6b63\u76f8\u5173\uff0cr=0\u8868\u793a\u4e24\u4e2a\u53d8\u91cf\u6ca1\u6709\u5173\u7cfb\uff0cr=-1\u8868\u793a\u4e24\u4e2a\u53d8\u91cf\u5448\u8d1f\u76f8\u5173\u3002\u5176\u5b9e\uff0c\u76f8\u5173\u7cfb\u6570\u77e9\u9635\u5c31\u662f\u6807\u51c6\u5316\u7684\u534f\u65b9\u5dee\u77e9\u9635\u3002\n\u4e0b\u56fe\u4e5f\u53ef\u4ee5\u770b\u5230\uff0c\u5404\u4e2a\u7279\u5f81\u4e0e\u6807\u7b7e\u7684\u76f8\u5173\u6027\u90fd\u8f83\u5f3a\uff0c\u6240\u4ee5\u6211\u4eec\u5728\u6b64\u7528\u4f8b\u4e0d\u7528\u4f5c\u7279\u5f81\u5254\u9664\u64cd\u4f5c\u3002"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "cols=['Clump_Thickness','Uniformity_of_Cell_Size','Uniformity_of_Cell_Shape','Marginal_Adhesion',\n                                         'Single_Epithelial_Cell_Size'\n                                        ,'Bare_Nuclei','Bland_Chromatin','Normal_Nucleoli','Mitoses','Class']\n\ncm = np.corrcoef(data[cols].values.T)\n#\u8bbe\u7f6e\u5b57\u7684\u6bd4\u4f8b\nplt.figure(figsize=(10,10))\nsns.set(font_scale=1.0)\n#\u7ed8\u5236\u76f8\u5173\u7cfb\u6570\u56fe\nhm = sns.heatmap(cm,cbar=True,annot=True,square=True,fmt=\".2f\",\n                 annot_kws={\"size\":10},yticklabels=cols,xticklabels=cols)\nplt.show()", "execution_count": 131, "outputs": [{"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f1461a400>", "image/png": "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\n"}, "metadata": {}}]}, {"metadata": {}, "cell_type": "markdown", "source": "# 2.KNN\u7b97\u6cd5\u4f7f\u7528\u793a\u4f8b\n\u90bb\u8fd1\u7b97\u6cd5\uff0c\u6216\u8005\u8bf4K\u6700\u8fd1\u90bb(kNN\uff0ck-NearestNeighbor)\u5206\u7c7b\u7b97\u6cd5\u662f\u6570\u636e\u6316\u6398\u5206\u7c7b\u6280\u672f\u4e2d\u6700\u7b80\u5355\u7684\u65b9\u6cd5\u4e4b\u4e00\u3002\u6240\u8c13K\u6700\u8fd1\u90bb\uff0c\u5c31\u662fk\u4e2a\u6700\u8fd1\u7684\u90bb\u5c45\u7684\u610f\u601d\uff0c\u8bf4\u7684\u662f\u6bcf\u4e2a\u6837\u672c\u90fd\u53ef\u4ee5\u7528\u5b83\u6700\u63a5\u8fd1\u7684k\u4e2a\u90bb\u5c45\u6765\u4ee3\u8868\u3002"}, {"metadata": {}, "cell_type": "markdown", "source": "## 2.1 \u8bad\u7ec3\u96c6\u6d4b\u8bd5\u96c6\u5212\u5206\n\u5c06\u8bad\u7ec3\u96c6\u6d4b\u8bd5\u96c6\u4f5c8\u6bd42\u5212\u5206"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "data_np=np.array(data)\n\n\ntrain_data, test_data= train_test_split(data_np,  test_size=0.2,random_state=2019) \n\ntrain_data_np=np.array(train_data)\ntest_data_np=np.array(test_data)\ntrain_label_np=train_data_np[:,-1]   #\u8bad\u7ec3\u96c6\u7684\u6807\u7b7e\ntest_label_np=train_data_np[:,-1]\n\ntrain_data_df= pd.DataFrame(train_data,columns=cols)\ntrain_data_df.to_csv('./train_data.csv',index=None)\n\ntest_data_df= pd.DataFrame(test_data,columns=cols)\ntest_data_df.to_csv('./test_data.csv',index=None)\ntrain_data_path='./train_data.csv'\ntest_data_path='./test_data.csv'", "execution_count": 238, "outputs": []}, {"metadata": {}, "cell_type": "markdown", "source": "## 2.2 \u8bad\u7ec3\u548c\u6d4b\u8bd5\n\u5982\u4e0bread_csv\u51fd\u6570\u53ef\u4ee5\u76f4\u63a5\u628acsv\u6587\u4ef6\u8f6c\u6362\u6210\u7b97\u6cd5\u6240\u9700\u8981\u7684\u683c\u5f0f\n\ncalculate_metric_value \u51fd\u6570\u53ef\u4ee5\u6d4b\u8bd5\u6a21\u578b\u6027\u80fd\uff0c\u8fd4\u56de\u5bf9\u5e94\u6570\u636e\u96c6\u5728\u8be5\u6a21\u578b\u4e0b\u9884\u6d4b\u7684f1\u503c\u3001\u51c6\u786e\u7387\u3001\u53ec\u56de\u7387\u3001\u67e5\u51c6\u7387\n\n\u8bad\u7ec3\u3001\u6d4b\u8bd5\u793a\u610f\u4ee3\u7801\u5982\u4e0b"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "from modelarts_pyspark.common import MeasurementType, Role, DataType, DfUtils, Meta, Attributes\nfrom pyspark.sql.functions import expr\nfrom pyspark.ml.evaluation import MulticlassClassificationEvaluator\nfrom pyspark.ml.feature import StringIndexer\ndef read_csv(path):\n\n    df = DfUtils.read_csv(sc, path, has_header=True)\n    new_list = DfUtils.modify_metadata(df.schema.fields, target='Class', name=Attributes.ROLE,\n                                       value=Role.TARGET)\n    fields = DfUtils.modify_metadata(new_list, target='Class', name=Attributes.MEASUREMENT,\n                                     value=MeasurementType.NOMINAL)\n    a=spark.createDataFrame(df.rdd, StructType(fields))\n    return spark.createDataFrame(df.rdd, StructType(fields))\n\n\ndef calculate_metric_value(knn_model, knn_df):\n    label_col=knn_df.select(\"Class\")\n    indexer = StringIndexer()\n    indexer.setInputCol('Class')\n    indexer.setOutputCol('label')\n    index_model = indexer.fit(label_col)\n    label_col = index_model.transform(label_col)\n    label_col=label_col.toPandas()\n    result = knn_model.transform(knn_df)\n\n    indexer = StringIndexer()\n    indexer.setInputCol('predict')\n    indexer.setOutputCol('predictIndexer')\n    index_model = indexer.fit(result)\n    index_result = index_model.transform(result)\n    index_result=index_result.toPandas()\n    dataset = pd.concat([index_result,label_col],axis=1)\n    dataset=spark.createDataFrame(dataset)\n    evaluator = MulticlassClassificationEvaluator(predictionCol=\"predictIndexer\")\n\n    metric_dict[\"f1\"] = (evaluator.evaluate(dataset, {evaluator.metricName: \"f1\"}))\n    metric_dict[\"accuracy\"] = (evaluator.evaluate(dataset, {evaluator.metricName: \"accuracy\"}))\n    metric_dict[\"precision\"] = (evaluator.evaluate(dataset, {evaluator.metricName: \"weightedPrecision\"}))\n    metric_dict[\"recall\"] = (evaluator.evaluate(dataset, {evaluator.metricName: \"weightedRecall\"}))\n    return index_result", "execution_count": 239, "outputs": []}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "from pyspark.sql.types import StructType\nfrom modelarts_pyspark import KNNClassification,KNNRegression\n\ntrain_data_frame = read_csv(train_data_path)\nknn_classification = KNNClassification().set_k(3)\nmodel = knn_classification.fit(train_data_frame)  #\u8bad\u7ec3\n\ntest_data_frame = read_csv(test_data_path)\nmetric_dict = {}\ncalculate_metric_value(model, test_data_frame)   #\u6d4b\u8bd5\nprint(metric_dict)", "execution_count": 240, "outputs": [{"output_type": "stream", "text": "{'f1': 0.9643627571057185, 'accuracy': 0.9642857142857143, 'precision': 0.9645453687092815, 'recall': 0.9642857142857143}\n", "name": "stdout"}]}, {"metadata": {}, "cell_type": "markdown", "source": "# 3.K\u6298\u4ea4\u53c9\u9a8c\u8bc1"}, {"metadata": {}, "cell_type": "markdown", "source": "K\u6298\u4ea4\u53c9\u9a8c\u8bc1(k-fold cross-validation)\u9996\u5148\u5c06\u6240\u6709\u6570\u636e\u5206\u5272\u6210K\u4efd\u5b50\u6837\u672c\uff0c\u4e0d\u91cd\u590d\u7684\u9009\u53d6\u5176\u4e2d\u4efd\u4e2a\u5b50\u6837\u672c\u4f5c\u4e3a\u6d4b\u8bd5\u96c6\uff0c\u5176\u4ed6K-1\u4efd\u6837\u672c\u7528\u6765\u8bad\u7ec3\u3002\u5171\u91cd\u590dK\u6b21\uff0c\u5e73\u5747K\u6b21\u7684\u7ed3\u679c\u6216\u8005\u4f7f\u7528\u5176\u5b83\u6307\u6807\uff0c\u6700\u7ec8\u5f97\u5230\u4e00\u4e2a\u5355\u4e00\u4f30\u6d4b\u3002\n\n\u8fd9\u4e2a\u65b9\u6cd5\u7684\u4f18\u52bf\u5728\u4e8e\uff0c\u4fdd\u8bc1\u6bcf\u4e2a\u5b50\u6837\u672c\u90fd\u53c2\u4e0e\u8bad\u7ec3\u4e14\u90fd\u88ab\u6d4b\u8bd5\uff0c\u964d\u4f4e\u6cdb\u5316\u8bef\u5dee\u3002K\u6298\u4ea4\u53c9\u9a8c\u8bc1\u5e38\u88ab\u7528\u4e8e\u6a21\u578b\u8c03\u4f18\uff0c\u627e\u5230\u4f7f\u5f97\u6a21\u578b\u6cdb\u5316\u6027\u80fd\u6700\u4f18\u7684\u8d85\u53c2\u503c\u3002\u627e\u5230\u540e\uff0c\u5728\u5168\u90e8\u8bad\u7ec3\u96c6\u4e0a\u91cd\u65b0\u8bad\u7ec3\u6a21\u578b\uff0c\u5e76\u4f7f\u7528\u72ec\u7acb\u6d4b\u8bd5\u96c6\u5bf9\u6a21\u578b\u6027\u80fd\u505a\u51fa\u6700\u7ec8\u8bc4\u4ef7\u3002\u5176\u4e2d\uff0c10\u6298\u30015\u6298\u4ea4\u53c9\u9a8c\u8bc1\u7b49\u6bd4\u8f83\u5e38\u7528\u3002\n\n\u672c\u4f8b\u4ee55\u6298\u4ea4\u53c9\u9a8c\u8bc1\u4e3a\u4f8b\u5b8c\u6210\u4e00\u4e2aK\u6298\u4ea4\u53c9\u9a8c\u8bc1\u5f97\u8fc7\u7a0b\u3002\u5c06\u4e0a\u97622.1\u4e2d\u5212\u5206\u7684\u8bad\u7ec3\u96c6\u4f5c\u4e3aK\u6298\u4ea4\u53c9\u9a8c\u8bc1\u7684\u6570\u636e\u96c6\uff0c\u5e76\u5728\u6b64\u4e0a\u5212\u5206\u8bad\u7ec3\u96c6\u3001\u9a8c\u8bc1\u96c6\u3002"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "from sklearn.model_selection import KFold\nmetric_dict = {}\nmodel_list=[]\ndef train_kfold(data,knn_k,k_fold_num,is_print_metric_dict=True):  #\u8f93\u5165\u4f9d\u6b21\u6570\u636e\uff0cknn\u91cc\u7684k\u503c\uff0cK\u6298\u4ea4\u53c9\u9a8c\u8bc1\u6298\u6570\uff0c\u662f\u5426\u6253\u5370\u6bcf\u6298\u6a21\u578b\u8bc4\u4f30\u4fe1\u606f\uff0c\u7136\u540e\u8fdb\u884c\u8bad\u7ec3\n    train_preds = np.zeros((len(data), 1))\n    result=[]\n    val_label_preds_list=[]\n    val_label_true_list=[]\n    fold = KFold(n_splits=k_fold_num, random_state=42)\n    for im, (trn_idx, val_idx) in enumerate(fold.split(data)):\n        train_data, val_data = data[trn_idx, :], data[val_idx, :]\n        train_data_df= pd.DataFrame(train_data,columns=cols)\n        train_data_df.to_csv('./train_data_k.csv',index=None)\n        val_data_df= pd.DataFrame(val_data,columns=cols)\n        val_data_df.to_csv('./val_data_k.csv',index=None)\n        train_data_path_k='./train_data_k.csv'\n        val_data_path_k='./val_data_k.csv'\n        train_data_frame = read_csv(train_data_path_k)\n        knn_classification = KNNClassification().set_k(knn_k)\n        model = knn_classification.fit(train_data_frame)\n        val_data_frame = read_csv(val_data_path_k)\n        val_label_preds=np.zeros((len(val_idx),1))\n        val_label_preds=np.array(calculate_metric_value(model, val_data_frame))[:,-1].reshape(len(val_idx), 1)\n        train_preds[val_idx]=val_label_preds\n        model_list.append(model)\n        if is_print_metric_dict:\n            print('knn_k:{},  fold_nums:{},  metric_dict:{}'.format(knn_k,im+1,metric_dict))\n    return train_preds", "execution_count": 241, "outputs": []}, {"metadata": {"trusted": true, "scrolled": true}, "cell_type": "code", "source": "train_preds=train_kfold(train_data,5,5)", "execution_count": 242, "outputs": [{"output_type": "stream", "text": "knn_k:5,  fold_nums:1,  metric_dict:{'f1': 1.0, 'accuracy': 1.0, 'precision': 1.0, 'recall': 1.0}\nknn_k:5,  fold_nums:2,  metric_dict:{'f1': 0.9289115646258501, 'accuracy': 0.9285714285714286, 'precision': 0.9298714430160336, 'recall': 0.9285714285714286}\nknn_k:5,  fold_nums:3,  metric_dict:{'f1': 0.9552191006272638, 'accuracy': 0.9553571428571429, 'precision': 0.9552491871570818, 'recall': 0.9553571428571428}\nknn_k:5,  fold_nums:4,  metric_dict:{'f1': 0.9644160151622838, 'accuracy': 0.9642857142857143, 'precision': 0.9651797477884434, 'recall': 0.9642857142857144}\nknn_k:5,  fold_nums:5,  metric_dict:{'f1': 0.9344834195898025, 'accuracy': 0.9369369369369369, 'precision': 0.9377949377949377, 'recall': 0.9369369369369369}\n", "name": "stdout"}]}, {"metadata": {}, "cell_type": "markdown", "source": "\u4e0a\u9762\u6253\u5370\u7684\u662f5\u6298\u4ea4\u53c9\u9a8c\u8bc1\u6bcf\u6b21\u8bad\u7ec3\u540e\u7528\u9a8c\u8bc1\u96c6\u8fdb\u884c\u9a8c\u8bc1\u7684\u8bc4\u4f30\u7ed3\u679c"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "from sklearn.metrics import confusion_matrix\nfrom sklearn import metrics\ndef calculate_kfold(label_np,train_preds):\n    label_np[label_np==2]=0.0\n    label_np[label_np==4]=1.0\n    cmat=confusion_matrix(label_np, train_preds)  #\u6df7\u6dc6\u77e9\u9635\n    fpr, tpr, thresholds=metrics.roc_curve(label_np,train_preds,pos_label=None,sample_weight=None,drop_intermediate=True)\n    plt.plot(fpr,tpr,marker = 'o')\n    plt.title(\"ROC\")\n    plt.show()\n    auc=metrics.auc(fpr,tpr)\n    print(\"auc\",auc)\n    accuracy = metrics.accuracy_score(label_np,train_preds) # \u8ba1\u7b97\u51c6\u786e\u7387\n    print (\"Accuary: %.2f%%\" % (accuracy * 100.0))\n    return accuracy,auc", "execution_count": 245, "outputs": []}, {"metadata": {}, "cell_type": "markdown", "source": "K\u6298\u4ea4\u53c9\u9a8c\u8bc1\u6bcf\u6b21\u5c06\u4e00\u90e8\u5206\u4f5c\u4e3a\u9a8c\u8bc1\u96c6\u3001\u5176\u4f59\u7684\u4f5c\u4e3a\u8bad\u7ec3\u96c6\uff0cK\u6298\u4ea4\u53c9\u9a8c\u8bc1\u4e4b\u540e\uff0c\u8bad\u7ec3\u96c6\u7684\u6bcf\u4e00\u90e8\u5206\u90fd\u66fe\u7ecf\u4f5c\u4e3a\u8fc7\u9a8c\u8bc1\u96c6\uff0c\u6240\u4ee5\u6bcf\u6b21\u5bf9\u9a8c\u8bc1\u96c6\u4f5c\u7684\u9884\u6d4b\u77e9\u9635\u62fc\u63a5\u8d77\u6765\u53ef\u4ee5\u5f97\u5230\u5bf9\u6240\u6709\u8bad\u7ec3\u96c6\u8fdb\u884c\u9884\u6d4b\u7684\u6807\u7b7e\uff0c\u5c06\u8fd9\u4e2a\u6807\u7b7e\u4e0e\u8bad\u7ec3\u96c6\u672c\u6765\u7684\u771f\u5b9e\u6807\u7b7e\u4f5c\u5bf9\u6bd4\u5f97\u5230\u8bc4\u4f30\u7ed3\u679c\u3002\n\nROC\u66f2\u7ebf\u3001auc\u503c\u4ee5\u53ca\u51c6\u786e\u7387\u6253\u5370\u5982\u4e0b\uff1a"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "accuracy,acc=calculate_kfold(train_label_np,train_preds)", "execution_count": 246, "outputs": [{"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f1c2089e8>", "image/png": "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\n"}, "metadata": {}}, {"output_type": "stream", "text": "auc 0.9510229342135216\nAccuary: 95.71%\n", "name": "stdout"}]}, {"metadata": {}, "cell_type": "markdown", "source": "# 4.\u7b97\u6cd5\u8c03\u4f18"}, {"metadata": {}, "cell_type": "markdown", "source": "## 4.1 \u5229\u7528\u4ea4\u53c9\u9a8c\u8bc1\u8fdb\u884c\u53c2\u6570\u5bfb\u4f18\n\u5c06KNN\u7684\u8d85\u53c2\u6570K\u503c\u5206\u522b\u8bbe\u7f6e\u4e3a1-10\uff0c\u7136\u540e\u5bf9\u5f53\u524d\u6a21\u578b\u505aK\u6298\u4ea4\u53c9\u9a8c\u8bc1\uff0c\u5206\u522b\u5c06KNN\u7684K\u503c\u4e3a1-10\u65f6\u7684ROC\u66f2\u7ebf\u3001auc\u503c\u3001\u51c6\u786e\u7387\u503c\u6253\u5370\u4e0b\u6765\uff1a"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "accuracy_list=[]\nauc_list=[]\nk_range=range(1,10)\nfor i in k_range:\n    train_preds=train_kfold(train_data,i,5,is_print_metric_dict=False)\n    accuracy,auc=calculate_kfold(train_label_np,train_preds)\n    accuracy_list.append(accuracy)\n    auc_list.append(auc)\n", "execution_count": 247, "outputs": [{"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f1405b2e8>", "image/png": "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\n"}, "metadata": {}}, {"output_type": "stream", "text": "auc 0.9495646483041202\nAccuary: 95.35%\n", "name": "stdout"}, {"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f26a0b7b8>", "image/png": "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\n"}, "metadata": {}}, {"output_type": "stream", "text": "auc 0.9235858183473707\nAccuary: 93.92%\n", "name": "stdout"}, {"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f1c1876d8>", "image/png": "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\n"}, "metadata": {}}, {"output_type": "stream", "text": "auc 0.9510229342135216\nAccuary: 95.71%\n", "name": "stdout"}, {"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f1c05f358>", "image/png": "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\n"}, "metadata": {}}, {"output_type": "stream", "text": "auc 0.9484051331664012\nAccuary: 95.53%\n", "name": "stdout"}, {"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f142f4ac8>", "image/png": "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\n"}, "metadata": {}}, {"output_type": "stream", "text": "auc 0.9510229342135216\nAccuary: 95.71%\n", "name": "stdout"}, {"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f142f48d0>", "image/png": "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\n"}, "metadata": {}}, {"output_type": "stream", "text": "auc 0.9510229342135216\nAccuary: 95.71%\n", "name": "stdout"}, {"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f7142f4a8>", "image/png": "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\n"}, "metadata": {}}, {"output_type": "stream", "text": "auc 0.9510229342135216\nAccuary: 95.71%\n", "name": "stdout"}, {"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f1c02a780>", "image/png": "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\n"}, "metadata": {}}, {"output_type": "stream", "text": "auc 0.9510229342135216\nAccuary: 95.71%\n", "name": "stdout"}, {"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f1c130c88>", "image/png": "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\n"}, "metadata": {}}, {"output_type": "stream", "text": "auc 0.9510229342135216\nAccuary: 95.71%\n", "name": "stdout"}]}, {"metadata": {}, "cell_type": "markdown", "source": "\u7ed8\u5236\u51faKNN\u7684\u8d85\u53c2\u6570K\u503c\u4e0e\u4ea4\u53c9\u9a8c\u8bc1\u5f97\u5230\u7684\u8bc4\u4f30\u7ed3\u679cauc\u503c\u548c\u51c6\u786e\u7387\u5173\u7cfb\u7684\u66f2\u7ebf\uff1a\n\u5982\u4e0b\u53ef\u4ee5\u770b\u5230\u5f53K\u503c\u53d63\u6216\u5927\u4e8e5\u7684\u65f6\u5019\uff0cauc\u548c\u51c6\u786e\u7387\u90fd\u6700\u9ad8\u3002\u4e3a\u4e86\u8282\u7701\u8ba1\u7b97\u91cf\uff0c\u6211\u4eec\u5c06K\u503c\u53d6\u4e3a3\u3001"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "plt.plot(k_range,accuracy_list,color='r',label='accuracy')\nplt.plot(k_range,auc_list,color='g',label='auc')\nplt.legend(loc=\"lower right\")\nplt.xlabel('k')\nplt.ylabel('auc or accuracy')\n", "execution_count": 249, "outputs": [{"output_type": "execute_result", "execution_count": 249, "data": {"text/plain": "Text(0,0.5,'auc or accuracy')"}, "metadata": {}}, {"output_type": "display_data", "data": {"text/plain": "<matplotlib.figure.Figure at 0x7f8f14272128>", "image/png": "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\n"}, "metadata": {}}]}, {"metadata": {}, "cell_type": "markdown", "source": "## 4.2 \u91cd\u65b0\u8bad\u7ec3\u6a21\u578b\n\u627e\u5230KNN\u6700\u4f18K\u503c\u4e4b\u540e\uff0c\u6211\u4eec\u5728\u5168\u90e8\u8bad\u7ec3\u96c6\u4e0a\u91cd\u65b0\u8bad\u7ec3\u6a21\u578b\uff0c\u5f97\u5230\u6700\u7ec8\u6a21\u578b\uff0c\u5e76\u4f7f\u7528\u72ec\u7acb\u6d4b\u8bd5\u96c6\u5bf9\u6a21\u578b\u6027\u80fd\u505a\u51fa\u6700\u7ec8\u8bc4\u4ef7\u3002"}, {"metadata": {"trusted": true}, "cell_type": "code", "source": "train_data_frame = read_csv(train_data_path)\nknn_classification = KNNClassification().set_k(3)\nmodel = knn_classification.fit(train_data_frame)  #\u8bad\u7ec3\n\ntest_data_frame = read_csv(test_data_path)\nmetric_dict = {}\ncalculate_metric_value(model, test_data_frame)   #\u6d4b\u8bd5\nprint(metric_dict)", "execution_count": 250, "outputs": [{"output_type": "stream", "text": "{'f1': 0.9643627571057185, 'accuracy': 0.9642857142857143, 'precision': 0.9645453687092815, 'recall': 0.9642857142857143}\n", "name": "stdout"}]}], "metadata": {"kernelspec": {"name": "pyspark-2.3.2", "display_name": "PySpark-2.3.2", "language": "python"}, "language_info": {"name": "python", "version": "3.6.4", "mimetype": "text/x-python", "codemirror_mode": {"name": "ipython", "version": 3}, "pygments_lexer": "ipython3", "nbconvert_exporter": "python", "file_extension": ".py"}}, "nbformat": 4, "nbformat_minor": 2}