Enron HamSpam.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "hide_input": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<script>\n",
       "code_show_err=false; \n",
       "function code_toggle_err() {\n",
       " if (code_show_err){\n",
       " $('div.output_stderr').hide();\n",
       " } else {\n",
       " $('div.output_stderr').show();\n",
       " }\n",
       " code_show_err = !code_show_err\n",
       "} \n",
       "$( document ).ready(code_toggle_err);\n",
       "</script>\n",
       "To toggle on/off output_stderr, click <a href=\"javascript:code_toggle_err()\">here</a>."
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import HTML\n",
    "HTML('''<script>\n",
    "code_show_err=false; \n",
    "function code_toggle_err() {\n",
    " if (code_show_err){\n",
    " $('div.output_stderr').hide();\n",
    " } else {\n",
    " $('div.output_stderr').show();\n",
    " }\n",
    " code_show_err = !code_show_err\n",
    "} \n",
    "$( document ).ready(code_toggle_err);\n",
    "</script>\n",
    "To toggle on/off output_stderr, click <a href=\"javascript:code_toggle_err()\">here</a>.''')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "hide_input": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>.prompt{width: 0px; min-width: 0px; visibility: collapse}</style>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(HTML('<style>.prompt{width: 0px; min-width: 0px; visibility: collapse}</style>'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "IPython.notebook.set_autosave_interval(60000)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Autosaving every 60 seconds\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.naive_bayes import MultinomialNB\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "import numpy as np\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn import metrics\n",
    "import seaborn as sn\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import itertools\n",
    "import email\n",
    "from string import punctuation\n",
    "import re\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "from nltk.corpus import stopwords\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn import tree\n",
    "stop_words = set(stopwords.words('english'))\n",
    "from sklearn.metrics import roc_curve,roc_auc_score\n",
    "from sklearn.utils import resample\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "%autosave 60"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "hide_input": false
   },
   "source": [
    "# Spam Email Classification with Scikit Learn\n",
    "\n",
    "### Introduction\n",
    "\n",
    "\n",
    "The Enron Spam Dataset is a collection of tens of thousands of emails which have been manually categorized into both Spam and Ham. There are two versions available - a preprocessed dataset and raw dataset. The preprocessed dataset contains the emails which have been converted into plain text format, and have been cleaned of HTML tags. The raw dataset contains the emails in their untouched form which contain the original HTML tags and various message formats, for example the '.eml' format. \n",
    "\n",
    "The goal of this project will be to train and evaluate different machine learning methods in classifying these emails as either ham or spam, and the Raw untouched dataset will be used in order to have full control of the process. Encorporating a cleaning step into the learning process will ensure that the model is robust in the classfication of new, unseen and 'dirty' data. This project will go through a number of steps before reaching a conclusion:\n",
    "\n",
    " - **Importing the data**\n",
    "     -  The dataset is split into two separate folders - Ham and Spam. In each folder there are various subfolders. An efficient loading process will be developed to read in each email into their respective categories. \n",
    "     \n",
    " - **Organizing the data**\n",
    "     -  Both Ham and Spam files will be combined into a pandas dataframe which will allow for simple manipulation cleaning stage, in the exploratory data stage and the model tuning stage. \n",
    "\n",
    " - **Cleaning the data**\n",
    "     -  Regular expressions functions will clear the email messages of html tags and other non alphanumeric noise. A lemmatizer will be applied to the emails in order to reduce words to their base format, for example: 'walking' will be converted to 'walk'.\n",
    "     \n",
    " - **Creating a train and test set**\n",
    "     -  The dataset will then be split into a training set and test set for use in model training. The models will be trained on the training set, and validated using the test data as 'unseen' data. \n",
    "     \n",
    " - **Feature Extraxtion with TfidfVectorizer**\n",
    "     -  The dataset will then be analysed for word frequencies. This converts the data into a numerical format which is required for the machine learning algorithms. \n",
    "     \n",
    " - **Model Training and Evaluation**\n",
    "     -  Model training begins at this step. Multiomial Naive Bayes, K Nearest Neighbour, Decision Tree, Random Forest and Logisitic Regression models will be trained in the training data, then validated on the test data. \n",
    "     \n",
    " - **Model Selection and Fine Tuning**\n",
    "     -  The best perfoming model from the previous training segment will be tunned via Gridsearch to find its optimum parameters.  \n",
    "\n",
    " - **Discussion**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Importing the data\n",
    "\n",
    "The raw dataset, as supplied, is separated into two categories: Ham and Spam. However the folder structure is quite disorganized and the emails are divided into various subdirectories. The file structure is also disorganied, with a mixture of text and .eml file types. Using the **OS** library, the output below shows that there are 249 folders in the Ham Directory, and 34 folders in the Spam Directory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 249 folders in the Ham Directory\n",
      "There are 34 folders in the Ham Directory\n"
     ]
    }
   ],
   "source": [
    "hamdir = r'C:\\Users\\Admin\\Downloads\\Compressed\\Ham Unprocessed'\n",
    "spamdir = r'C:\\Users\\Admin\\Downloads\\Compressed\\Spam Unprocessed'\n",
    "dirs=[hamdir,spamdir]\n",
    "\n",
    "def count_dirs(dirs):\n",
    "    for d in dirs:\n",
    "        folders = 0\n",
    "        for _, dirnames, filenames in os.walk(d):\n",
    "          # ^ this  means \"we won't be using this value\"\n",
    "            folders += len(dirnames)\n",
    "        print('There are',folders,'folders in the Ham Directory')\n",
    "\n",
    "count_dirs(dirs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In order to capture all files in the folder structure, quite a convoluted piece of code will be used. **Os.walk** will search through each subdirectory and append the files to a pre-defined list called *mylist*. However, since many of these files have different formats, the library **email** will be used. This will allow us to extract the body of the email message. \n",
    "\n",
    "In some cases, there is more than one level in an email ie. more than one 'body'. In the **email** library, the body of an email is known as a 'payload'. Due to the multi level emails, the extraction function has to check twice if the message body or payload is a multipart before appending the data to the list. It first checks if the message is multipart. If it is not, the message is saved. If it is a multipart email, the functions goes one level deeper and checks again if the message is multipart ie. one of two messages. If it is not, the data is saved, if it is multipart, the code can extract both messages. \n",
    "\n",
    "Once it finally has the payloads, this is appended to the list and iterates again over the next file in the directory. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "hide_input": false
   },
   "outputs": [],
   "source": [
    "def getlist(directory):\n",
    "    # this function takes the directory name, iterates over each file subdirectories\n",
    "    # and appends them to the list 'mylist'\n",
    "    mylist=[]\n",
    "    for directory, subdirectory, filenames in  os.walk(directory):\n",
    "        for filename in filenames:\n",
    "            with open(os.path.join(directory, filename), \"r\",encoding=\"latin-1\") as f:\n",
    "                data = f.read()\n",
    "             \n",
    "            # the below code checks for multi part emails and appends them to 'mylist'\n",
    "            b = email.message_from_string(data)\n",
    "            if b.is_multipart():\n",
    "                for payload in b.get_payload():\n",
    "                    if payload.is_multipart():\n",
    "                        for payload1 in payload.get_payload():\n",
    "                            mylist.append(payload1.get_payload())\n",
    "                    else:        \n",
    "                        mylist.append(payload.get_payload())\n",
    "            else:\n",
    "                mylist.append(b.get_payload())\n",
    "\n",
    "    return mylist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# sending the ham and spam directories to the getlist function\n",
    "# this will return a list of emails \n",
    "\n",
    "hamlist=getlist(hamdir)\n",
    "spamlist=getlist(spamdir)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Organizing the data\n",
    "\n",
    "Once both Ham and Spam lists have been created, they will be combined into a pandas dataframe which will allow for simple manipulation during cleaning stage, the exploratory data stage and the model tuning stage. FIrst, both the ham and spam lists are converted into a Pandas dataframe, giving them the column ['email'] and['target']. Target is the label variable, with 0  being ham and 1 being spam. \n",
    "\n",
    "Both dataframes are then concatenated to form then main dataframe called **'all_emails'** which will be used throughout this process."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_df(hamlist,spamlist):\n",
    "    \n",
    "    # takes both ham and spam lists and concatenated them into a dataframe\n",
    "    # returns dataframe once complete\n",
    "\n",
    "    # create dataframe of ham\n",
    "    ham=pd.DataFrame(hamlist,columns =['email'])\n",
    "    # add target column for ham\n",
    "    ham['target']=0\n",
    "    \n",
    "    # do same process for spam\n",
    "    spam=pd.DataFrame(spamlist,columns =['email'])\n",
    "    spam['target']=1\n",
    "    #concatenate ham and spam and shuffle dataframe\n",
    "    all_emails=pd.concat([ham,spam])\n",
    "    all_emails = all_emails.sample(frac=1).reset_index(drop=True)\n",
    "\n",
    "    \n",
    "    return all_emails"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# send the newly created ham list and spam list to the create_df function\n",
    "# this will return a new dataframe which will be saved as 'all_emails'\n",
    "\n",
    "all_emails = create_df(hamlist,spamlist)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cleaning the Data\n",
    "\n",
    "Regular expressions suitable for these email files were created and will be used on each message. URLs, HTML tags, email addresses, sequences of digits and other non-alphanumeric noise will be removed with a **clean_regex** function. A sufficiently tidy set of emails should remain, however we can go two steps further by removing stop words and lemmatizing the dataset after the regex function.\n",
    "\n",
    "Lemmatizing is a natural language processing technique which essentially truncates words to their base form. This allows for simplifying of the dataset without losing much meaning. For example, a string with the words 'walk, walks, walked, walking' will, after lemmatization, contain only instances of the word walk. There is another similar method of cleaning text called Stemming, however I decided againt this method as there is potential for the data to end up containing words that don't exist, for example, ponies could become poni, as a stemmer simply removes the plural segment 'ies'. Lemmatizing would convert it to pony. For the lemmatization process I will be using **WordNetLemmatizer** from the nltk.stem library \n",
    "\n",
    "Stop words such as **'the, a, an, in'** are words of little importance and unneccesary for this type of natural language processing task. These are more than likely not the keywords that will signify whether an email is spam or ham, and are more useful for sentiment analysis. Thefore I will remove them from the collection of emails which will reduce the dimensional space of the dataset, and increase overall importance of the word space. For this process I will use the **stopwords** package from the nltk.corpus library. \n",
    "\n",
    "Before the cleaning process, we will check a single row for noise such as punctuation, HTML tags or URLs and stop words. This can then be compared to the same email after the cleaning process.  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "data": {
      "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>email</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Louise and John,\\n\\nThank you for having the m...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>&lt;html&gt;\\n\\n&lt;head&gt;\\n&lt;meta http-equiv=3D\"Content-...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>&lt;HTML&gt;&lt;HEAD&gt;&lt;META HTTP-EQUIV=3D\"Content-Type\" ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Due to recent events, mixed messages are perme...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>FYI \\n\\nadr\\n---------------------- Forwarded ...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               email  target\n",
       "0  Louise and John,\\n\\nThank you for having the m...       0\n",
       "1  <html>\\n\\n<head>\\n<meta http-equiv=3D\"Content-...       1\n",
       "2  <HTML><HEAD><META HTTP-EQUIV=3D\"Content-Type\" ...       1\n",
       "3  Due to recent events, mixed messages are perme...       0\n",
       "4  FYI \\n\\nadr\\n---------------------- Forwarded ...       0"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_emails.head(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For the cleaning process I will write two functions - **clean_regex** and **clean_column**, which are contained in the two code blocks below. The dataframe to be cleaned is sent to the **clean_column** function, which works on the 'email' column of the dataset. It first applys the **clean_regex** function on each item in the 'email' column, and then moves on to stop word removal. Finally, two types of lemmatization are applied: verb and noun lemmatization, signified by the argument **pos=v** or **pos=n**."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def clean_regex(m):\n",
    "        \n",
    "        #defining the cleaning function which will run several regex search & replace\n",
    "        # processes on each email\n",
    "        # m for message in email column\n",
    "        m = re.sub(r'^[a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\\.[a-zA-Z0-9-.]+$','',str(m)) # email addresses\n",
    "        m = re.sub(r'(http|https|ftp)://[a-zA-Z0-9\\\\./]+', ' ', str(m)) # http/url regex\n",
    "        m = re.sub(r'\\d+', '', str(m)) # numbers\n",
    "        m = re.sub(r'<[^<]+?>', '', str(m)) # html <tags>\n",
    "        m = m.replace(r'[^a-zA-Z]', '') # non alphanumerics\n",
    "        m = m.replace('nbsp', '') # common in unprocessed spam files, new html\n",
    "        m = m.translate(str.maketrans('', '', punctuation)) # remove punctuation\n",
    "        m = m.lower() # lower case\n",
    "\n",
    "        return m\n",
    "    \n",
    "def clean_column(df,col_name):\n",
    "    \n",
    "    # apply clean regex to column\n",
    "    df[col_name]=df[col_name].apply(clean_regex)\n",
    "    \n",
    "    # apply stop word removal to column\n",
    "    df[col_name]=df[col_name].apply(lambda x: ' '.join([item for item in x.split() if item not in stop_words]))\n",
    "    \n",
    "    # keep only words between 3 and 15 characters long (some very long gibberish strings in some emails)\n",
    "    df[col_name]=df[col_name].apply(lambda x: ' '.join([item for item in x.split() if 3 <= len(item) <= 15]))\n",
    "    \n",
    "    # initialize Lemmatizer\n",
    "    lem = WordNetLemmatizer()\n",
    "    \n",
    "    # apply verb lemmatizer\n",
    "    df[col_name]=df[col_name].apply(lambda x: ' '.join([lem.lemmatize(word,pos='v') for word in x.split()]))\n",
    "\n",
    "    # apply noun lemmatizer\n",
    "    df[col_name]=df[col_name].apply(lambda x: ' '.join([lem.lemmatize(word,pos='n') for word in x.split()]))\n",
    "    \n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# send all_emails dataset to clean_column function\n",
    "# arguments are:  dataset, column to clean\n",
    "\n",
    "all_emails_clean = clean_column(all_emails,'email')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "hide_input": true
   },
   "source": [
    "The whole column of emails should now be cleaned, so the same row as before can be printed to observe the results. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "data": {
      "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>email</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>louise john thank meet gas power origination m...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>notification limit account access notification...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>hey hear new drmicrog call copyiaacutelis thin...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>due recent event mix message permeate appropri...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>fyi adr forward audrey sharon solon ricki audr...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               email  target\n",
       "0  louise john thank meet gas power origination m...       0\n",
       "1  notification limit account access notification...       1\n",
       "2  hey hear new drmicrog call copyiaacutelis thin...       1\n",
       "3  due recent event mix message permeate appropri...       0\n",
       "4  fyi adr forward audrey sharon solon ricki audr...       0"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_emails_clean.head(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Imbalanced Classes - Downsampling\n",
    "\n",
    "The dataset must be split into two segments: a training set and test set for use in model training. The models will be trained on the training set, and validated using the test data as 'unseen' data.\n",
    "\n",
    "However, before the data is split, it must be balanced due to spam comprising the larger class majority by almost a factor of 2. There are roughly 40,000 spam emails, and 19,000 ham emails. Ideally the dataset should be balanced so as to not create bias in the models during training. A high bias such as that observed in the current dataset could cause some models to predict more spam than ham in an effort to reach a higher accuracy level, and while a high accuracy level may mean that many spam emails are being classified correctly, it may also cause unwanted false positives where the model predicts a high number of ham as spam. \n",
    "\n",
    "I will create a function called **downsamp_maj** to downsample the majority class (spam) to the same level as the minority (ham), which is roughly 19,000 observations, and then proceed with the train test split. Even though we are removing close to 20,000 spam emails, we are still left with a large number for training so overall it is not a drastic reduction in observations. Ideally, we would just search for more ham examples to balance the dataset naturally. \n",
    "\n",
    "Before the downsampling begins, we can quickly check the proportion of spam to ham, where spam==1 and ham==0:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 41032 spam emails and 19088 ham emails in the dataset\n"
     ]
    }
   ],
   "source": [
    "# check value counts before sample \n",
    "s=all_emails.target.value_counts()\n",
    "\n",
    "print('There are',s.values[0],'spam emails and',s.values[1],'ham emails in the dataset')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It is a highly imbalanced dataset as observed in the printout. The majority class will now be downsampled in the code below using the **resample** package from from sklearn.utils and the proportion of classes will be checked once again when the downsampling is completed. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def downsamp_maj(df,target):\n",
    "\n",
    "    # this function can be used for downsampling the majority class to match the minority class \n",
    "    # this function works for any panda dataframe with two classes labeled 0 and 1\n",
    "    # arguents: panadas dataframe and response variable column name as string\n",
    "    \n",
    "    # get counts of both classes\n",
    "    lenclass_1 = len(df[df[target]==1])\n",
    "    lenclass_0 = len(df[df[target]==0])\n",
    "\n",
    "    # if class 1 is larger than class 0\n",
    "    if lenclass_1 > lenclass_0:\n",
    "\n",
    "        # set class 1 as the majority and class 0 as minority\n",
    "        df_majority = df[df[target]==1]\n",
    "        df_minority = df[df[target]==0]\n",
    "        \n",
    "        # downsample majority to the number of classes in minority\n",
    "        df_majority_downsampled = resample(df_majority, \n",
    "                                 replace=False,    # sample without replacement\n",
    "                                 n_samples=lenclass_0,     # to match minority class\n",
    "                                 random_state=123) # reproducible results \n",
    "\n",
    "        # join downsampled majortiy and minority into dataframe\n",
    "        df = pd.concat([df_majority_downsampled, df_minority],ignore_index=True)\n",
    "        \n",
    "    # otherwise downsample class 0\n",
    "    else:\n",
    "\n",
    "        df_majority = df[df[target]==0]\n",
    "        df_minority = df[df[target]==1]\n",
    "\n",
    "        df_majority_downsampled = resample(df_majority, \n",
    "                                 replace=False,    # sample without replacement\n",
    "                                 n_samples=lenclass_1,     # to match minority class\n",
    "                                 random_state=123) # reproducible results \n",
    "\n",
    "        # Combine minority class with downsampled majority class\n",
    "        df = pd.concat([df_majority_downsampled, df_minority],ignore_index=True)\n",
    "\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 19088 spam emails and 19088 ham emails in the dataset\n"
     ]
    }
   ],
   "source": [
    "# send the email dataset to the function downsamp_maj which takes the dataframe, and the name of the column\n",
    "# containing the class labels, in this case the name is 'target'\n",
    "\n",
    "balanced_emails = downsamp_maj(all_emails,'target')\n",
    "\n",
    "# get value counts for printout\n",
    "s=balanced_emails.target.value_counts()\n",
    "print('There are',s.values[0],'spam emails and',s.values[1],'ham emails in the dataset')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Exploratory Data Analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Belopw are two plots of the top 20 most frequent words by class. Surprisingly, company is the top Spam word. This word seems out of place in a spam dataset but this will be re-weighted in the feature extraction stage with the TFID Vectorizer. And not surpisingly, enron is the most frequent word in the ham dataset, followed by company. The word 'company' occurs very frequently in both datasets, and may be a missed html or email tag."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "top_spam = pd.Series(' '.join(balanced_emails['email'][balanced_emails['target']==1]).split()).value_counts()[:20]\n",
    "top_spam.plot(kind='bar')\n",
    "plt.title('Top 20 Most Frequent Words in Spam Emails')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "top_ham = pd.Series(' '.join(balanced_emails['email'][balanced_emails['target']==0]).split()).value_counts()[:20]\n",
    "top_ham.plot(kind='bar')\n",
    "plt.title('Top 20 Most Frequent Words in Ham Emails')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train Test Split\n",
    "\n",
    "The train test split process can now begin. In order to asses the performance of varios models, a training set and test set is needed. The train set will be used to train the models, and then the models will be evaluated on the test set. The split proportion is usally 80/20 train/test or 70/30 train test. For this project I will be splitting the data 70/30 ie. 70% of the data will be used for training, and 30% for evaluation.\n",
    "\n",
    "The function below randomly separates the dataframe and returns four pandas series - xtrain, xtest, ytrain and ytest. \n",
    "\n",
    "**xtrain:** These are the predictor variables, in this case just one variable which is the collection of emails. And since it is the training set, it contains 70% of the emails. **ytrain:** This is the correspinding class labels [0,1] for the emails in xtrain. **xtest:** These are the emails used for validation - 30% split. **ytest:** These are the corresponding class labels [0,1] for the emails in our test set above. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "def tt_split(df):\n",
    "    \n",
    "    xtrain, xtest, ytrain, ytest = train_test_split(df['email'], df['target'], test_size=0.3,random_state=5)\n",
    "    \n",
    "    return xtrain,xtest,ytrain,ytest\n",
    "\n",
    "xtrain,xtest,ytrain,ytest = tt_split(balanced_emails)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Extraction with TFID Transformer\n",
    "\n",
    "In order to train the models with the email data, it will need to be converted to numerical format. One useful method is to transform the data using sklearn's TFID Transformer. TFID stands for *'Term Frequency Inverse Document'*. Term Frequency records the number of times a word appears in a document to the total number of words in that document. The more times a word increases in each document, the higher the ratio or term frequency of that word. Each document has its own term frequency ratio. \n",
    "\n",
    "After recording the Term Frequency, the inverse data frequency process is used to calculate the weight of words across all documents in the dataset. Words that rarely occur have a high 'IDF' score. The IDF of common words will approach zero, which shows they are not significant. For example, if the word 'the' is present in every email, it is of no use to learning algorithms, therefore will have a low IDF score. Words which appear less commonly will have a higher IDF. The Term Frequency and IDF are then combined to give the TFIDF score. This process adjusts the weight of words which in turn adjusts their significance.\n",
    "\n",
    "In the below code block I transform the data using the sklearn TfidfVectorizer. The n_gram argument of TfidfVectorizer also checks for pairs or triplets of words, for example, 'very expensive' instead of just the single words 'very' and 'expensive'. This paring of words may give more insight into the document context. In this instance n_grams=2 is chosen. Performing this tranformation results in two new vectors which can be used for training and testing: **xtrain_tfid** and **xtest_tfid.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "def tfid_vec(train,test):\n",
    "    \n",
    "    # initialize vectorizer - we will check ngrams up to length 2\n",
    "    transformer = TfidfVectorizer(ngram_range=(1,2),min_df=5)\n",
    "    \n",
    "    # using the vectorizer to transform the training and test data to numerical values.  \n",
    "    train_tfid = transformer.fit_transform(train)\n",
    "    test_tfid = transformer.transform(test)\n",
    "    \n",
    "    return train_tfid,test_tfid\n",
    "\n",
    "xtrain_tfid,xtest_tfid = tfid_vec(xtrain,xtest)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model Training and Evaluation\n",
    "\n",
    "Various models will now be trained:: Multiomial Naive Bayes, K Nearest Neighbour, Decision Tree, Random Forest and Logisitic Regression models will each take the xtrain subset along with the class labels of that subset: ytrain. Each model will be initialized with their default settings, and then evaluated by recording various metrics such as precision, confusion matrix and accuracy. The best performing model will be isolated, then optimized to get the maxmimum possible accuracy metric via hyperparameter tuning using GridSearch. First I will initialize and train the models on the training data **xtrain.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Admin\\Anaconda3\\lib\\site-packages\\sklearn\\ensemble\\forest.py:246: FutureWarning: The default value of n_estimators will change from 10 in version 0.20 to 100 in 0.22.\n",
      "  \"10 in version 0.20 to 100 in 0.22.\", FutureWarning)\n",
      "C:\\Users\\Admin\\Anaconda3\\lib\\site-packages\\sklearn\\linear_model\\logistic.py:433: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n"
     ]
    }
   ],
   "source": [
    "# Multinomial Naive Bayes\n",
    "mymodelNB = MultinomialNB().fit(xtrain_tfid, ytrain)\n",
    "\n",
    "#K Nearest Neightbours\n",
    "mymodelKNN = KNeighborsClassifier().fit(xtrain_tfid,ytrain)\n",
    "\n",
    "#Decision Tree\n",
    "mymodelDT= tree.DecisionTreeClassifier().fit(xtrain_tfid,ytrain)\n",
    "\n",
    "# Random Forest\n",
    "mymodelRF = RandomForestClassifier().fit(xtrain_tfid, ytrain)\n",
    "\n",
    "# Logistic Regression\n",
    "mymodelLR = LogisticRegression().fit(xtrain_tfid,ytrain)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To evaluate the models I will create two functions. The first is **plot_matrix** which creates a visually pleasing confusion matrix plot and takes as arguments the test labels, the predicted classes from the models, the class names and the name of the model. This will then be called in the second function **evaluate_models** which will first predict which class the 'unseen' test emails belong to using the model.predict function, then computes the accuracy score, classification report and ROC curve. The arguments for this function are: the model, test emails and test labels. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_matrix(ytest,ypreds,classes,title):\n",
    "    # Code adapted and tweaked from scikit-learn.org\n",
    "\n",
    "    cm=confusion_matrix(ytest, ypreds,labels=[1,0])\n",
    "  \n",
    "    plt.imshow(cm, interpolation='nearest', cmap=plt.cm.Blues)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=45)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "\n",
    "    fmt = 'd'\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, format(cm[i, j], fmt),\n",
    "                 horizontalalignment=\"center\",\n",
    "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "    plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate_model(model,xtest,ytest):\n",
    "    \n",
    "    # get class predictions\n",
    "    ypreds=model.predict(xtest)\n",
    "    \n",
    "    # get class probabilities for use in ROC curve\n",
    "    yprobs=model.predict_proba(xtest)[:,1]\n",
    "    \n",
    "    # set model name\n",
    "    model_name = type(model).__name__\n",
    "    \n",
    "    print('\\n---------',model_name,'---------\\n')\n",
    "    \n",
    "    # accuracy score\n",
    "    score = metrics.accuracy_score(ytest, ypreds, normalize=True, sample_weight=None)\n",
    "    print('\\n\\nAcruacy score of ',score)\n",
    "    \n",
    "    # call plot_matrix function \n",
    "    classes=['spam','ham']\n",
    "    plt.figure()\n",
    "    plot_matrix(ytest,ypreds,classes,model_name)\n",
    "    \n",
    "    # classification report\n",
    "\n",
    "    print(metrics.classification_report(ytest, ypreds,target_names=['Spam','Ham']))\n",
    "    \n",
    "    # Compute false positive rate, true positive rate, thresholds and area under the ROC curve\n",
    "    fpr, tpr, thresholds = roc_curve(ytest, yprobs)\n",
    "    roc_auc = roc_auc_score(ytest, yprobs)\n",
    "    \n",
    "    plt.figure()\n",
    "    \n",
    "    # Plot ROC curve\n",
    "    plt.plot(fpr, tpr, label='ROC curve (area = %0.3f)' % roc_auc)\n",
    "    plt.legend(loc=\"best\")\n",
    "    plt.plot([0, 1], [0, 1], 'k--')  # random predictions curve\n",
    "    plt.xlim([0.0, 1.0])\n",
    "    plt.ylim([0.0, 1.0])\n",
    "    plt.xlabel('False Positive Rate or (1 - Specifity)')\n",
    "    plt.ylabel('True Positive Rate or (Sensitivity)')\n",
    "    plt.title('ROC '+model_name)\n",
    "    \n",
    "    return score,model_name"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "hide_input": false
   },
   "source": [
    "Now that the functions have been set up, each trained model can be passed into the **evaluate_model** function, with accuracy recorded in a results list."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--------- MultinomialNB ---------\n",
      "\n",
      "\n",
      "\n",
      "Acruacy score of  0.9359993014930585\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        Spam       0.89      0.99      0.94      5712\n",
      "         Ham       0.99      0.88      0.93      5741\n",
      "\n",
      "   micro avg       0.94      0.94      0.94     11453\n",
      "   macro avg       0.94      0.94      0.94     11453\n",
      "weighted avg       0.94      0.94      0.94     11453\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Multinomial Naive Bayes Performance Metrics\n",
    "\n",
    "results=[]\n",
    "\n",
    "score,model_name=evaluate_model(mymodelNB,xtest_tfid,ytest)\n",
    "\n",
    "results.append((model_name,score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--------- KNeighborsClassifier ---------\n",
      "\n",
      "\n",
      "\n",
      "Acruacy score of  0.6342443028027591\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        Spam       1.00      0.27      0.42      5712\n",
      "         Ham       0.58      1.00      0.73      5741\n",
      "\n",
      "   micro avg       0.63      0.63      0.63     11453\n",
      "   macro avg       0.79      0.63      0.58     11453\n",
      "weighted avg       0.79      0.63      0.58     11453\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#K Nearest Neighbours Performance Metrics\n",
    "\n",
    "score, model_name = evaluate_model(mymodelKNN,xtest_tfid,ytest)\n",
    "\n",
    "results.append((model_name,score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--------- DecisionTreeClassifier ---------\n",
      "\n",
      "\n",
      "\n",
      "Acruacy score of  0.9551209290142321\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        Spam       0.95      0.96      0.96      5712\n",
      "         Ham       0.96      0.95      0.96      5741\n",
      "\n",
      "   micro avg       0.96      0.96      0.96     11453\n",
      "   macro avg       0.96      0.96      0.96     11453\n",
      "weighted avg       0.96      0.96      0.96     11453\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Decision Tree Performance Metrics\n",
    "\n",
    "score,model_name = evaluate_model(mymodelDT,xtest_tfid,ytest)\n",
    "\n",
    "results.append((model_name,score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--------- RandomForestClassifier ---------\n",
      "\n",
      "\n",
      "\n",
      "Acruacy score of  0.9615821182222999\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        Spam       0.95      0.98      0.96      5712\n",
      "         Ham       0.98      0.95      0.96      5741\n",
      "\n",
      "   micro avg       0.96      0.96      0.96     11453\n",
      "   macro avg       0.96      0.96      0.96     11453\n",
      "weighted avg       0.96      0.96      0.96     11453\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Random Forest Performance Metrics**\n",
    "\n",
    "score,model_name = evaluate_model(mymodelRF,xtest_tfid,ytest)\n",
    "\n",
    "results.append((model_name,score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--------- LogisticRegression ---------\n",
      "\n",
      "\n",
      "\n",
      "Acruacy score of  0.9797432986990309\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        Spam       0.99      0.97      0.98      5712\n",
      "         Ham       0.97      0.99      0.98      5741\n",
      "\n",
      "   micro avg       0.98      0.98      0.98     11453\n",
      "   macro avg       0.98      0.98      0.98     11453\n",
      "weighted avg       0.98      0.98      0.98     11453\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUgAAAEYCAYAAAA+mm/EAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzt3Xm8XdP9//HX+95MSCJBYiaGIGhpEEGrpBpDTa35qxFDBT9aqt/W0MFQNK222hhqKEX7bVFjiooIVVSIeSYJQQgRMhBJZPj8/tjrpsd1Jtxzz3Dfzz72456z9tp7r31v87HWXmuvpYjAzMw+qanaBTAzq1UOkGZmBThAmpkV4ABpZlaAA6SZWQEOkGZmBThAdhCSDpZ052c89llJO7RxkWqOpIsl/bTa5bDaIY+DrE2SpgDfiYi72vm6VwJTI+Inn+HYAD4EApgNXAv8MCIWt2khzdqJa5DW1jaLiO7AV4EDgMPb+gLK+P+7VnH+P1mdkXSkpEmS3pM0WtJqOfuGSnpR0mxJF0m6V9J30r5DJd2fPkvSeZKmp7xPSdpU0gjgYOBHkj6Q9I+Uf4qkndLnZkmnSpos6X1Jj0pas3U5I2IS8ACweU75lpd0uaRpkt6QdJak5pzz/kbSDEmvSDpOUkjqlPb/S9LZkh4gq6WuW+J866f7n53OeW2xe0/7rpR0Vpm/65B0tKSJkmZKulCS2uJvbLXDAbKOSBoC/ALYH1gVeBW4Ju1bCbgeOAVYEXgR2LbAqYYC2wMbAL3IanrvRsSlwP8Bv4qI7hGxR55jTwQOAnYDepLVED/MU9aNgK8Ak3KSrwIWAesDX0rl+E7adySwK1lAHQjsnefaw4ARQI9078XO93PgTqA3sAZwfrF7z1P+gr/rHLsDWwGbpXw75ymz1TEHyPpyMHBFRDwWEQvIguE2kvqRBaxnI+LGiFgEjALeKnCehWRBZiOy59DPR8S0MsvwHeAnEfFiZJ6MiNwA85ikucDzwL+AiwAkrUwWAE+IiLkRMR04DzgwHbc/8PuImBoRM4GRea59ZUQ8m+5vhRLnWwisDawWEfMj4v5Pee/FftctRkbErIh4DbiHnNqyNQYHyPqyGllNBoCI+ICs9rN62vd6zr4ApuY7SUTcDVwAXAi8LelSST3LLMOawOQi+wcC3clqZlsDy6X0tYHOwDRJsyTNAi4B+ubc2+s558n9nC+t1Pl+BAh4OPXCHw6f6t6L/a5b5P4H6MN039ZAHCDry5tkgQEAScuRNaffAKaRNSVb9in3e2sRMSoitgA2IWtu/rBlV4kyvA6sVyxDqlleBzwI/CznuAXAShHRK209I2KTtP9j5ScLxJ84datyFDxfRLwVEUdGxGrAUcBFktYvce+5iv2urYNwgKxtnSV1a9mA64DDJG0uqStwDvBQREwBbgO+IGnv1LFxLLBKvpNK2krS1pI6A3OB+UDLUJy3gXWLlOmPwM8l9U8dHl+UtGKBvCOBEZJWSc3YO4HfSOopqUnSepK+mvJeBxwvaXVJvYCTiv1iSp1P0n6SWgLuTLLgurjEvef6K4V/19ZBOEDWttuBeTnbV4CfAjeQ1bjWIz1zi4gZwH7Ar8iaghsDj5DVslrrCVxGFjheTfl/nfZdDmycmq035zn2t2TB7E5gTsq/TL7CR8TTwL38t4Z2CNAFeC5d+3qyDhBSee4EngIeT/e+iPzBq0Wx820FPCTpA2A0cHxEvFLi3nPLPo4Cv2vrODxQvEEpGyc4FTg4Iu6pdnk+LUm7AhdHxNolM5tViGuQDUTSzpJ6pSbhqWSdFOOrXKyySFpG0m6SOklaHTgNuKna5bKOzQGysWxD1sM8A9gD2Dsi5lW3SGUTcAZZ0/dxsmFCPyt6hFmFuYltZlaAa5BmZgV0qnYBKkmdlgl16VHtYlgrXxqwVrWLYHk89tijMyKiT1udr7nn2hGLSj/hiXnvjImIXdrqum2psQNklx503XD/ahfDWnngoQuqXQTLY5nOerV0rvLFonll/fub/8SFK7XlddtSQwdIM6siCZqaq12Kz8UB0swqp86n7XSANLPKqfMpMh0gzaxC5BqkmVlews8gzczyk5vYZmYFuYltZlaAa5BmZnl4HKSZWRFuYpuZ5eNhPmZmhTX5GaSZ2Sd5HKSZWSFuYpuZFeZhPmZmBbgGaWaWh8dBmpkV4Sa2mVk+7qQxMyuszmuQ9R3ezax2SdDUqfRW1qk0RdLTkp6Q9EhKW0HSWEkT08/eKV2SRkmaJOkpSQNzzjM85Z8oaXip6zpAmlnlSKW38u0YEZtHxJbp+8nAuIjoD4xL3wF2BfqnbQTwh6woWgE4DdgaGASc1hJUC3GANLPKUVPp7bPbC7gqfb4K2Dsn/erIjAd6SVoV2BkYGxHvRcRMYCxQdD1uB0gzq4yWYT6ltvIEcKekRyWNSGkrR8Q0gPSzb0pfHXg959ipKa1QekHupDGzyimvCb1Sy3PF5NKIuLRVnu0i4k1JfYGxkl4odtU8aVEkvSAHSDOrGJUXIGfkPFfMKyLeTD+nS7qJ7Bni25JWjYhpqQk9PWWfCqyZc/gawJspfYdW6f8qdl03sc2sIkQWIEttJc8jLSepR8tnYCjwDDAaaOmJHg7ckj6PBg5JvdmDgdmpCT4GGCqpd+qcGZrSCnIN0swqQ0JtMx/kysBNKZh2Av4aEXdImgBcJ+kI4DVgv5T/dmA3YBLwIXAYQES8J+nnwISU78yIeK/YhR0gzaxiymxiFxURLwOb5Ul/F/hanvQAji1wriuAK8q9tgOkmVVMWwTIanKANLOKcYA0M8tDbfcMsmocIM2sYlyDNDMrwAHSzKwAB0gzs3yEn0GameUjyntTppY5QJpZxThAmpkVUt/x0QHSzCpE0NRU3/PhOECaWcW4iW1mloc7aexzeeG2M3h/7gIWL1nCosVL+PLBv+LPIw+jf7+VAejVYxlmvT+PwQeOBOB/Dx/KoXttw+IlS/jBr67nrgefp2uXTtx1+Ql06dKJTs3N3HTX45x18e3VvK2GNH/+fHbacXs+WrCARYsX8c1v7ctPTzuDe+4ex6kn/ZAlS5awXPfuXHb5lay3/vrVLm7tqO/46ABZbbuM+D3vzpq79Puwk/+09PPIE7/J7A/mAbDRuquw384DGbjv2azaZ3luv/g4vrD3mSz4aBG7jBjF3Hkf0alTE3dfcSJ3PvAcDz89pb1vpaF17dqVO8beTffu3Vm4cCFDvvplhu68K9877hj+fsMtbDRgAJf84SJGnnMWl11xZbWLWxsa4BlkfZe+we3z9YFcd8ejAOy+wxf5+5jH+GjhIl59810mvz6DrTbtB8DceR8B0LlTM506NZNNh2dtSRLdu3cHYOHChSxauHDpjNhz5swBYM6c2ay62mrVLGbNaYsZxavJNcgqigj+cdFxRASX3/AAV9z4wNJ92w1cj7ffe5/Jr70DwOp9luehnFrhG9Nnslrf5QFoahL/+etJrLdmHy659t9MeObVdr2PjmLx4sVsO2gLJk+exFHHHMugrbfmokv+yDf33I1uyyxDz549uff+8dUuZm2p7fhXkmuQVTTksPPY9n9+yd7HXcRRB3yF7Qaut3Tf/rtsyd/vyFnoLc9/aVsqikuWBIMPHMn6O/+ELTddm43XW7XSRe+QmpubeejRJ5g0ZSqPTHiYZ595hvN/fx43jb6dyVOmMmz4YZz0vydWu5g1QxJNTU0lt1pW26VrcNPemQ3AOzM/YPTdT7HVJv0AaG5uYq8hm3H9mMeW5n1j+izWWKX30u+r9+299PgWsz+Yx78fmcjQbTeufOE7sF69erH9V3dgzJh/8vRTTzJo660B2He/Axg//j9VLl1tqfcmdsUCZFqJ7DZJT0p6RtIBkqZI+qWkh9O2fsq7h6SHJD0u6S5JK6f00yVdJenOdOy3JP1K0tOS7pDUuVLlr7Rlu3Wh+7Jdl37eaZuNeHbymwAM2XpDXpryNm9Mn7U0/23/eor9dh5Il86dWHu1FVl/rT5MeGYKK/XuzvLdlwGgW9fODNl6Q16c8nb731CDe+edd5g1K/t7zJs3j7vH3cVGGw1gzuzZTHzpJQDuvmssG240oJrFrDn1HiAr+QxyF+DNiPgGgKTlgV8CcyJikKRDgN8BuwP3A4MjIiR9B/gR8IN0nvWAHYGNgQeBfSLiR2lt3G8AN+deVNIIYAQAnbtX8PY+n74r9uDa3x4JQKfmZq795yOM/c/zAOy38xZLO2daPP/yW9xw5+M8fsOPWbR4CSeMvI4lS4JVVurJZWcOo7mpiaYmccPYx/jnfc+0+/00uremTePIw4ezePFilsQS9tl3f3b7xu5cePFlHLT/PjQ1NdGrd28uuazs9aA6htqOfyWpUj2ekjYgW3P2OuDWiLhP0hRgSES8nGp/b0XEipK+APwGWBXoArwSEbtIOh1YGBFnS2oC5gHdUiA9E3gvIn5XqAxNy/aNrhvuX5H7s89u5oQLql0Ey2OZzno0IrZsq/N1XaV/rHHwqJL5Xv7tbm163bZUsSZ2RLwEbAE8DfxC0s9aduVmSz/PBy6IiC8ARwHdcvIsSOdbQhYsW45ZgnvhzWqWyPoWS221rJLPIFcDPoyIvwC/BgamXQfk/HwwfV4eeCN9Hl6pMplZeyr9/LEjP4P8AnCupCXAQuAY4Hqgq6SHyILzQSnv6cDfJb0BjAfWqWC5zKyd1Hj8K6liATIixpA9g1wq/dfiwog4o1XeW4Bb8pzj9FbfuxfaZ2Y1RtlLDPXMz/DMrCJE/QfIdh0oHhH9ImJGe17TzKqnLTtpJDWnsdK3pu/rpPHTEyVdK6lLSu+avk9K+/vlnOOUlP6ipJ1LXdNv0phZxbRxJ83xwPM5338JnBcR/YGZwBEp/QhgZkSsD5yX8iFpY+BAYBOycdoXSWoudkEHSDOrCKVnkKW28s6lNcheDPlj+i5gCFnHL8BVwN7p817pO2n/11L+vYBrImJBRLwCTAIGFbuuA6SZVUjZw3xWkvRIzjYiz8l+R/aG3ZL0fUVgVkQsSt+nAqunz6sDrwOk/bNT/qXpeY7Jy500ZlYxZbagZxR7k0bS7sD0iHhU0g4tyXmyRol9xY7JywHSzCqmjQaCbwfsKWk3srfsepLVKHtJ6pRqiWsAb6b8U4E1gamSOpG9iPJeTnqL3GPychPbzCqirZ5BRsQpEbFGRPQj62S5OyIOBu4B9k3ZhvPfsdSj+e8befum/JHSD0y93OsA/YGHi13bNUgzq5gKv0lzEnCNpLOAx4HLU/rlwJ8lTSKrOR4IEBHPSroOeA5YBBwbEYuLXcAB0swqpq3ftY6IfwH/Sp9fJk8vdETMB/YrcPzZwNnlXs8B0swqxu9im5nl0fIMsp45QJpZhdT+dGalOECaWcXUeXx0gDSzCnET28wsv2zJBQdIM7O8HCDNzAqo8/joAGlmFeJnkGZm+cnDfMzMCqvz+OgAaWaV01TnEbJggJTUs9iBETGn7YtjZo2i0V81fJZPzsLb8j2AtSpYLjNrAHUeHwsHyIhYs9A+M7Ny1HsnTVkziks6UNKp6fMakraobLHMrBG05brY1VAyQEq6ANgRGJaSPgQurmShzKz+CWiWSm61rJxe7G0jYqCkxwEi4j1JXSpcLjOrd+oY4yAXSmoiLY8oaUX+uzatmVlBdR4fy3oGeSFwA9BH0hnA/cAvK1oqM6t7IhsHWWqrZSVrkBFxtaRHgZ1S0n4R8Uxli2VmjaCRx0HmagYWkjWzvZa2mZVUD73UpZTTi/1j4G/AasAawF8lnVLpgplZ/Wv4JjbwbWCLiPgQQNLZwKPALypZMDOrf7Ud/korJ0C+2ipfJ+DlyhTHzBqFgOZGfQYp6TyyZ44fAs9KGpO+DyXryTYzK6zBx0G29FQ/C9yWkz6+csUxs0ZS5/Gx6GQVl7dnQcyssbRVE1tSN+DfQFeymHV9RJwmaR3gGmAF4DFgWER8JKkrcDWwBfAucEBETEnnOgU4AlgMfC8ixhS7djm92OtJukbSU5Jeatk+682aWceh1MwutpVhATAkIjYDNgd2kTSY7IWV8yKiPzCTLPCRfs6MiPWB81I+JG0MHAhsAuwCXCSpudiFyxnTeCXwJ7L/IOwKXEcWtc3MilIZWymR+SB97Zy2AIYA16f0q4C90+e90nfS/q8pi8R7AddExIKIeAWYBAwqdu1yAuSyLdXQiJgcET8hm93HzKwgqexxkCtJeiRnG/HJc6lZ0hPAdGAsMBmYFRGLUpapwOrp8+rA6wBp/2xgxdz0PMfkVc4wnwUp+k6WdDTwBtC3jOPMrIMr81XDGRGxZbEMEbEY2FxSL+AmYEC+bOlnvou2Xh2h9TF5lVOD/D7QHfgesB1wJHB4GceZWQfX1hPmRsQs4F/AYKCXpJZK3hrAm+nzVGDN7PrqBCwPvJebnueYvEoGyIh4KCLej4jXImJYROwZEQ+Uf0tm1hGJ0s3rcl41lNQn1RyRtAzZxDnPA/cA+6Zsw4Fb0ufR6Ttp/90RESn9QEldUw94f+DhYtcuNlD8JopUPyPiWyXuq+o2G7AW9z4wqtrFsFZ6f+XkahfB2kPbTVaxKnBV6nFuAq6LiFslPQdcI+ks4HGgZWji5cCfJU0iqzkeCBARz0q6DngOWAQcm5ruBRV7BnnB57kjM7O2WFIhIp4CvpQn/WXy9EJHxHxgvwLnOhs4u9xrFxsoPq7ck5iZtSbqf1XDcueDNDP71Op8rgoHSDOrnA4TICV1jYgFlSyMmTUOqf6nOyvnXexBkp4GJqbvm0k6v+IlM7O619bjINtbOQPFRwG7k82KQUQ8iV81NLMSOsSqhkBTRLzaqjeq6NghMzOo/xX+ygmQr0saBEQaqPldwNOdmVlRkur+GWQ5AfIYsmb2WsDbwF0pzcysqBpvQZdUMkBGxHTSqzpmZp9GnVcgSwdISZeR553siPjEnG1mZi1aOmnqWTlN7LtyPncDvsnHJ500M/skQXOd99KU08S+Nve7pD+TzehrZlaUylpUoXZ9llcN1wHWbuuCmFljyZrY1S7F51POM8iZ/PcZZBPZ/Gqe0M/MSmroYT5pLZrNyNahAViSZuY1MyuqEWqQRR+hpmB4U0QsTpuDo5mVp4z3sGu9k7ucPqaHJQ2seEnMrOE07LvYkjqlNWW/DBwpaTIwl6zmHBHhoGlmBYnGHubzMDAQ2LudymJmDUU0NfAwHwFExOR2KouZNZBsTZpql+LzKRYg+0g6sdDOiPhtBcpjZo1C9d+LXSxANgPdoc7ryGZWFdkzyPoOH8UC5LSIOLPdSmJmDafWe6lLKfkM0szss6rz+Fg0QH6t3UphZg1HNPCSCxHxXnsWxMwajOq/iV3vAd7MalRbrWooaU1J90h6XtKzko5P6StIGitpYvrZO6VL0ihJkyQ9lfsmoKThKf9EScNLXdsB0swqRmVsZVgE/CAiBgCDgWMlbUw2q9i4iOgPjOO/s4ztCvRP2wjgD5AFVOA0YGtgEHBaS1AtxAHSzCqmLSariIhpEfFY+vw+8DywOrAXcFXKdhX/fetvL+DqyIwHeklaFdgZGBsR70XETLKJv3cpdu3PMmGumVlJQjSX9wxyJUmP5Hy/NCIuzXtOqR/wJeAhYOWImAZZEJXUN2VbnY8vCzM1pRVKL8gB0swqRuUFyBkRsWUZ5+oO3ACcEBFzipw7344okl6Qm9hmVjFt9AwSSZ3JguP/RcSNKfnt1HQm/Zye0qcCa+YcvgbwZpH0ghwgzawylNUgS20lT5Nluhx4vtUcEKOBlp7o4cAtOemHpN7swcDs1BQfAwyV1Dt1zgxNaQW5iW1mFSEo9xlkKdsBw4CnJT2R0k4FRgLXSToCeA3YL+27HdgNmAR8CBwG2dhuST8HJqR8Z5Ya7+0AaWYV0xbhMSLuL3KqT7zxl5aGObbAua4Arij32g6QZlYxdf4ijQOkmVVG9i52fUdIB0gzq5DaX5SrFAdIM6uYOo+PDpBmVhluYpuZFSJoqvOR1nVe/MZx7FFHsN5aqzB4iy8uTXv6qSfZ6avbsc2Wm3HAPnsyZ86cjx3z+muvsdpKPRl13m/au7gN7YUbT2LCX05g/FXf4/4rjgPgx0fsxOTRpzD+qu8x/qrvsfM2GwKwQs9lueOCI3ln3Bmc94M9P3ae/b++GRP+cgIP//l4bjnvMFZcftl2v5dqUxn/q2UOkDXif4YN54Zbbv9Y2nePGcHpZ53Dg488ye577s2o8379sf2n/OhEdhpadDIS+4x2OfZSBg8fxZcPv2Bp2vnX3M/g4aMYPHwUYx58EYD5Hy3kzEvv5JQLPv63a25u4twT9mCXYy9l0LDf88yktzh6323b9R6qLZsPsvRWyxwga8R2X96e3ius8LG0SRNfZLsvbw/AjkO+zuibb1y679bRN9NvnXUZsPEm7VpO+7gP5y/kP0+9yvwFiz6W3rIm9HLLdAGgx3JdmTZjTp4zNDbXIK1iBmy8KbffOhqAm2+8njemZjM1zZ07l9/95lxO/vHPqlm8hhUR/OP3R/DAn47j8L0GLU0/et9tefjPx3Pxj/elV49lip5j0eIlHH/uzUz4ywm8/I9TGdBvZa78x4SixzSitphRvJraNUBK6ifpmfa8Zj278JI/ctklF7H9tlvxwQfv07lLVhs55+en8/++ezzdu3evcgkb05Cj/sC2h57P3if+iaP22YbtNl+Hy24cz8b7/oqtDxnFWzPmMPJ73yh6jk7NTRz5rcEMHj6Kdfc4h2cmT+OHh+zYTndQGxqhie1e7Bq2wYYbcfOt2WQjkya+xJh/Zs+5Hp3wMKNvuoHTfnwys2fPQk1NdOvWjRHH5H391D6laTPeB+CdmXMZfe+zbLXxGjzwxCtL919xywRu/HXx5Uw222A1AF55I5sL4fpxT/O/w75aoRLXqtpvQpdSjQDZLOkyYFvgDbLp0b9NtnZEF7IZOIZFxIeSrgTmARsBa5PNyjEc2AZ4KCIObffSt6N3pk+nT9++LFmyhHNHns3hR44A4I5x9y7N84uzzmC55bo7OLaRZbt1pqlJfPDhRyzbrTM7bd2fc64Yxyor9uCtd7PAudcOm/Dcy28XPc+b78xmo359WanXcsyYNZevDVqfF6e80x63UDvKXFKhllUjQPYHDoqIIyVdB+wD3BgRlwFIOgs4Ajg/5e8NDAH2BP5BNvXRd4AJkjaPiCdaX6AeHX7I/3D/fffy7owZDFhvLU756WnM/WAul11yEQB77PVNvn3IYVUuZePru0IPrh05DMiaydfe+QRjx7/E5T/bny9usBoRwavTZvLdX9609JgXbjyJHst1pUunZvbYfhN2P/5yXpgynXOuGMfYPxzFwkWLee2tWYz4+d+rdVtV0YbTnVWNspmB2uli2XoSY9MqZEg6CegM3AecBfQCugNjIuLoVIMcGxH/J2ndlN5y7NVkgfXmVtcYQVYbZc0119rimZdewWrLyjueWu0iWB7zx//y0XKWPijXgC98Kf500z0l823Tv3ebXrctVaMXe0HO58VktdgrgeMi4gvAGUC3PPmXtDp2CXlqwBFxaURsGRFbrtinT1uW28w+rbZac6FKamWYTw9gWlp34uBqF8bM2ka9j4OslV7sn5It4/gq8DRZwDSzOlfrw3hKadcAGRFTgE1zvue+O/eHPPkPLXLsoa3zm1mNcYA0M/uk7BFjfUdIB0gzqwyPgzQzK8wB0swsr9rvpS7FAdLMKsY1SDOzPOpgHHhJDpBmVjGq8yqkA6SZVUydx8eaedXQzBpQW7yKLekKSdNzJ9uWtIKksZImpp+9U7okjZI0SdJTkgbmHDM85Z8oqfiEnokDpJlVhrImdqmtDFcCrVenOxkYl2b3Gpe+A+xKNqVif7JZvf4AWUAFTgO2BgYBp7UE1WIcIM2sIloWLiu1lRIR/wbea5W8F3BV+nwVsHdO+tWRGQ/0krQqsDPZ1InvRcRMYCyfDLqf4GeQZlYxZT6CXEnSIznfL42IS0scs3JETAOIiGmS+qb01YHXc/JNTWmF0otygDSzyikvQs5owwlz810xiqQX5Sa2mVVMBZd9fTs1nUk/p6f0qcCaOfnWAN4skl68/J+1dGZmpVRwQvHRZAv4kX7ekpN+SOrNHgzMTk3xMcBQSb1T58zQlFaUm9hmVjltMA5S0t+AHcieVU4l640eCVwn6QjgNWC/lP12YDey1VE/JFsJlYh4T9LPgQkp35kR0brj5xMcIM2sItpqPsiIOKjArq/lyRtA3jWQI+IK4IpPc20HSDOrDHnJBTOzwhwgzczy8XyQZmYF1ftkFQ6QZlYRLa8a1jMHSDOrGDexzcwKcA3SzKyAOo+PDpBmViHykgtmZnm5k8bMrIg6j48OkGZWOa5BmpkV4GeQZmYF1Hd4dIA0swopd1GuWuYAaWYV4ya2mVkB9R0eHSDNrILqvALpAGlmleL5IM3M8vKbNGZmRThAmpkV4Ca2mVk+HgdpZpafn0GamRXhJraZWQGuQZqZFVDn8dEB0swqp97fxVZEVLsMFSPpHeDVapejjawEzKh2IewTGunvsnZE9Gmrk0m6g+z3U8qMiNilra7blho6QDYSSY9ExJbVLod9nP8uja2p2gUwM6tVDpBmZgU4QNaPS6tdAMvLf5cG5meQZmYFuAZpZlaAA6SZWQEOkGZmBThAmpkV4ABpZlaAA6TZ56R6f+HYCvIwnzomaSdgN+AW4IWIeLvKRerQJB0I9AeuAV6LiAVVLpJ9Tq5B1ilJmwHnkP0NjwCGSVq7uqXqWHJrjpIOAH4ArAWcC+wmqUe1ymZtwwGyDklahaym8uuIOIGsxrIqsJ+kdapauA5CkiI1vyT1BboDB0fEkcCdwJ7A1yT1rGIx7XNyE7vOSNoFuAh4C+gUEYNy0vcEXgfOi4j51StlY2sVHI8HjiGrbNwfEYen9KOBrwN/Am4L/0OrS65B1hFJGwGHAd8EdgC6SboGICLuAG4FbnZwrKyc4LgdsDUwBDgA2EDSWSnPxcDtwGMOjvXLNcg6kJ519QJ+RvaPcUREPJT2PQxMj4jdq1jEDiX9PdYHzgeWAMMi4l1JG5PV7h+PiO9Xs4zWNlzKCAy+AAAGIElEQVSDrAORmQlcBtwP7Czpi2nfIGAtSQM93KRycn+36e8xEfg9MB8YKmnFiHgO+C4wQFIf/z3qn2uQNU7SUGBH4BXgZqA3cBTwHnBrRDxRxeJ1CK2eOR4CrAK8QNaE3hk4mOzxxp0RMUNSl4j4qGoFtjbjGmQNk7Q78AvgCWB34BJgHnAhsBqwt6Tukvx3rKCc4HgCcDgwi2xIz1nAPcDVwEHAjpKaHBwbh/9h1ShJvYFdgP3IguIqwEtkwXEecB5wbUR8EBFLqlbQBpb7Hx5JGwCbATsBPclWNF0OOI0sSP4e+I//Fo3FTewalf5x9iFrUv8ZOBAIYDQwDdjDvdXtI40tfRvoS1ZzH0nWWfZt4CTgpog4tXoltErxutg1RtL2ZIO+P4qImyR1I3uNcLKkrYFxwGUOjpUjaVtgrYi4RtJ3gePJaon/Ias53h8RiyQtBG4DRlWvtFZJDpA1RNJWwB/JOmO2lXRARBwoqZ+kv5DVWg6PiGeqWtDG1xv4RRp3ugZZR8wQsqE9ywAnSOqT0r8eEW9VraRWUW5i1whJXyF73vjPiPhnSnsQuIOsSTcQmOde6/Yh6evAb4HxEXGkpK5kf58+wDrAv4GHI+K1KhbTKsydNDVA0rrAPsBwslpKi2FA/4hYEBEPOji2n4gYC/wE2EvSgWlmnr8Bc8iGWN3t4Nj43MSuMkl7AqcD3wCeA74v6T7gSWBtskHHKwAz/cpa+4qIWyQtImtuk55J/gnoHhFzql0+qzw3satI0ubAlcBBEfF8SvsL2XCS+4Fm4PaIuLlqhTQk7Uq2/vX3I+L6apfH2o8DZBVJGkA2TORBYGWyCSjeIKs5rgkcHRFjJDVHxOKqFdRanklOjoiXq10Waz8OkFUkqTtwKNlbGL8hGwi+PTAR2Ag4BRjqXmuz6nCArAEt7+5K2pLstbVjI+KeNAbv9oiYXOUimnVIDpA1QFIzsDnZVFnnRMQtVS6SmeEAWTMkLQf0jYhXWqbJcq+1WXU5QJqZFeCB4mZmBThAmpkV4ABpZlaAA6SZWQEOkB2ApMWSnpD0jKS/S1r2c5xrB0m3ps97Sjq5SN5ekv7fZ7jG6ZL+t9z0VnmulLTvp7hWP0keiG95OUB2DPMiYvOI2BT4CDg6d6cyn/r/CxExOiJGFsnSC/jUAdKsVjhAdjz3AeunmtPzki4CHgPWlDRU0oOSHks1ze4AknaR9IKk+4FvtZxI0qGSLkifV5Z0k6Qn07Yt2TyW66Xa67kp3w8lTZD0lKQzcs71Y0kvSroL2LDUTUg6Mp3nSUk3tKoV7yTpPkkvpYXPkNQs6dycax/1eX+R1vgcIDsQSZ2AXYGnU9KGwNUR8SVgLtn8hztFxEDgEeDEtOTDZcAewFfIFg/LZxRwb0RsRja577PAyWQTPGweET9MS9j2BwaRvTm0haTtJW1BtubOl8gC8FZl3M6NEbFVut7zwBE5+/oBXyWbQu7idA9HALMjYqt0/iPTWjNmBXk+yI5hGUktk+3eB1xOtvjUqxExPqUPBjYGHkgv8nQhm2VoI+CViJgIS6djG5HnGkOAQwDSzEOz08qMuYam7fH0vTtZwOxBtvDVh+kao8u4p00lnUXWjO8OjMnZd11aXXCipJfTPQwFvpjzfHL5dO2XyriWdVAOkB3DvIjYPDchBcG5uUnA2Ig4qFW+zclWU2wLAn4REZe0usYJn+EaVwJ7R8STkg4lmyquRetzRbr2dyMiN5Aiqd+nvK51IG5iW4vxwHaS1geQtGxaC/oFYB1J66V8BxU4fhxwTDq2WVJP4H2y2mGLMcDhOc82V5fUl2x9l29KWkZSD7LmfCk9gGmSOgMHt9q3n6SmVOZ1gRfTtY9J+ZG0QXr/3awg1yANgIh4J9XE/pYWqAL4SUS8JGkEcJukGWQznW+a5xTHA5dKOgJYDBwTEQ9KeiANo/lneg45AHgw1WA/AL4dEY9JuhZ4AniV7DFAKT8FHkr5n+bjgfhF4F6ySYiPjoj5kv5I9mzysTQZyDvA3uX9dqyj8mQVZmYFuIltZlaAA6SZWQEOkGZmBThAmpkV4ABpZlaAA6SZWQEOkGZmBfx//4be3PqMeEgAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Logistic Regression Performance Metrics\n",
    "\n",
    "score,model_name = evaluate_model(mymodelLR,xtest_tfid,ytest)\n",
    "\n",
    "results.append((model_name,score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "hide_input": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MultinomialNB 0.936\n",
      "KNeighborsClassifier 0.6342\n",
      "DecisionTreeClassifier 0.9551\n",
      "RandomForestClassifier 0.9616\n",
      "LogisticRegression 0.9797\n"
     ]
    }
   ],
   "source": [
    "for name,score in results:\n",
    "    print(name,round(score,4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Selection and Fine Tuning\n",
    "\n",
    "The above list of accuracy scores shows that **Logistic Regression** is the highest performing model in terms of accuracy rate. Indeed it also has a very good AUC (Area under ROC curve). The AUC tells us the predicitve power of the model and how well it can distinguish between classes. An AUC of 1 is a perfect result, and the Logistic Regression model is very close with a low false negative rate. The precision score is also very high, which is the proportion of positive (spam) predicitons that were actually correct. Second best is the Multinomial Naive Bayes model, however it has a lower precision and higher false negative rate. Although the performance metrics of the Logistic Regression model are very good, one last hyperparameter grid search will be carried out to see if it is in any way possible to increase the performance metrics even further. \n",
    "\n",
    "**'C' parameter:** A regularization function. A smaller number for C specifies a stronger regularization. Default is 1. Regularization can reduce the problem of overfitting, however if if C too large, the model my overfit on the data and won't generalize well on test data. If C is too small, the data complexity will be lowered, or over-regularized, which will cause underfit. For this grid search I will test the regularization factor from 0.001 to 1000 in steps of 10.\n",
    "\n",
    "**'Penalty' parameter:** Two loss functions: L1 and L2. L2 minimizes sum of squares error, and L1 minimizes absolute error. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Admin\\Anaconda3\\lib\\site-packages\\sklearn\\linear_model\\logistic.py:433: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best hpyerparameters :  {'C': 1000, 'penalty': 'l2'}\n",
      "Accuracy : 0.9881749803540022\n"
     ]
    }
   ],
   "source": [
    "grid_params = {\"C\":[0.001, 0.01, 0.1, 1, 10, 100, 1000], \"penalty\":[\"l1\",\"l2\"]}\n",
    "log_grid  = LogisticRegression()\n",
    "mymodelLR_Grid = GridSearchCV(log_grid,param_grid=grid_params,cv=10,n_jobs=-1)\n",
    "mymodelLR_Grid = mymodelLR_Grid.fit(xtrain_tfid,ytrain)\n",
    "print(\"Best hpyerparameters : \",mymodelLR_Grid.best_params_)\n",
    "print(\"Accuracy :\",mymodelLR_Grid.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The values C = 1000 and Penalty = L2 were chosen as the best parameters. This new logistic regression model will now be used on the test data to ascertain if the performance has increased. It will be initialized as 'best_LR', and sent to the  **evaluate_model** function."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--------- LogisticRegression ---------\n",
      "\n",
      "\n",
      "\n",
      "Acruacy score of  0.9873395616868943\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        Spam       0.99      0.98      0.99      5712\n",
      "         Ham       0.98      0.99      0.99      5741\n",
      "\n",
      "   micro avg       0.99      0.99      0.99     11453\n",
      "   macro avg       0.99      0.99      0.99     11453\n",
      "weighted avg       0.99      0.99      0.99     11453\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "best_LR = mymodelLR_Grid.best_estimator_\n",
    "score,model_name = evaluate_model(best_LR,xtest_tfid,ytest)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As we can see, these new parameters have boosted the accuracy score of the Logistic Regression model. The model can now be saved using the pickle library for later use."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "\n",
    "# save the model to disk\n",
    "\n",
    "filename = 'Best_Model_LogiRegression.sav'\n",
    "pickle.dump(best_LR, open(filename, 'wb'))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Discussion\n",
    "\n",
    "From the various models created in this project, the Logistic Regression model came out on top, followed closely by Multinomial Naive Bayes. It might be difficult to improve this model more - perhaps more training data would be useful to increase the accuracy rate. Roughly 20,000 'ham' emails were removed to balance the dataset. It is yet to be seen whether upsampling the 'spam' emails would improve or worsen the results. \n",
    "\n",
    "For real world use, it is difficult to say if this model would be suitable. The dataset came from an American corporate environment, specifically the Enron corporation which closed in 2007. There are some unanswered questions that are outside of the scope of this project. Would the model generalize well to data from emails outside of the corporate world? Or even emails outside of America? Would it work on emails ten years later in the the year 2019? Personal 'ham' emails would have more casual langauge, in the year 2007 or present day. This model has also been tuned on the English language only. As a general spam filter, this may work OK in an English speaking corporate environment, but more data is needed from other sources to truly say if it can generalize well as a spam filter.\n",
    "\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.1"
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "oldHeight": 495,
   "position": {
    "height": "466px",
    "left": "1469px",
    "right": "20px",
    "top": "107px",
    "width": "353px"
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "varInspector_section_display": "block",
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}