python_files/case_study_classification_bins_coulmn_difference.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Table of Contents\n",
    "\n",
    "* [Case study classification project](#casestudy)\n",
    "* [EDA - exploratory data analysis - get to know the data](#eda)\n",
    "* [Get the sweet summary stats on each column to understand the distribution better](#summary)\n",
    "* [Get a  representation of the numbers](#numbersrep)\n",
    "* [Outliers](#outliers)\n",
    "* [Box Plots](#boxplots)\n",
    "* [Outliers](#outliers)\n",
    "* [Pre-processing](#preprocessing)\n",
    "* [Split off the dependant variable (label) ](#split)\n",
    "* [TT split and get model](#ttsplit)\n",
    "* [Apply model, train model](#applymoded)\n",
    "* [Evaluate accuracy and test](#evaluate)\n",
    "* [Visualising accuracy - ROC / AUC](#rocauc)\n",
    "* [Data imbalanced -  SMOTE (oversample the minority class of the target label)](#smote)\n",
    "\n",
    "\n",
    "* [sample](#sample)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Case study classification project <a class=\"anchor\" id=\"casestudy\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import libraries\n",
    "import pandas as pd \n",
    "import seaborn as sns \n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np \n",
    "from scipy import stats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# conda install -c conda-forge jupyter_contrib_nbextensions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# df = pd.read_csv('creditcardmarketing.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_original = pd.read_excel('creditcardmarketing.xlsx')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save as a csv file to use in the SQL task\n",
    "#import csv\n",
    "#df_original.to_csv(r'new_creditcardmarketing_original.csv', index = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# create a copy of the dataaet\n",
    "df = df_original"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###  EDA - exploratory data analysis - get to know the data <a class=\"anchor\" id=\"eda\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# check information about the dataset\n",
    "#df.dtypes\n",
    "#df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# cheack shape\n",
    "#df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# print the sum of null values \n",
    "#print(df.isnull().sum())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# standize the readers: replace spaces by _, put in lowercase, remove #\n",
    "df.columns = df.columns.str.replace(' ', '_')\n",
    "df.columns= df.columns.str.lower()\n",
    "df.columns= df.columns.str.replace('#_', '')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# find the percentage of missing values\n",
    "#pct_missing=24/18000\n",
    "#pct_missing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b>We found few missing values (less than 20%). so we are not droping them.<br/><b>\n",
    "Since the balances columns are float, we will fill them with the mean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # fill nulls with the means\n",
    "\n",
    "# col_miss_values= ['average_balance', 'q1_balance', 'q2_balance', 'q3_balance', 'q4_balance']\n",
    "\n",
    "# mean_average_balance = df['average_balance'].mean()\n",
    "# df['average_balance'] = df['average_balance'].fillna(mean_average_balance)\n",
    "\n",
    "# mean_average_balance = df['q1_balance'].mean()\n",
    "# df['q1_balance'] = df['q1_balance'].fillna(mean_average_balance)\n",
    "\n",
    "# mean_q2_balance = df['q2_balance'].mean()\n",
    "# df['q2_balance'] = df['q2_balance'].fillna(mean_q2_balance)\n",
    "\n",
    "# mean_q3_balance = df['q3_balance'].mean()\n",
    "# df['q3_balance'] = df['q3_balance'].fillna(mean_q3_balance)\n",
    "\n",
    "# mean_q4_balance = df['q4_balance'].mean()\n",
    "# df['q4_balance'] = df['q4_balance'].fillna(mean_q4_balance)\n",
    "                                           \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "col_miss_values= [ 'q2_balance', 'q3_balance', 'q4_balance']\n",
    "\n",
    "\n",
    "mean_q2_balance = df['q2_balance'].mean()\n",
    "df['q2_balance'] = df['q2_balance'].fillna(mean_q2_balance)\n",
    "df['q2_balance'] = pd.to_numeric(df['q2_balance'], errors = 'coerce')\n",
    "\n",
    "mean_q3_balance = df['q3_balance'].mean()\n",
    "df['q3_balance'] = df['q3_balance'].fillna(mean_q3_balance)\n",
    "df['q3_balance'] = pd.to_numeric(df['q3_balance'], errors = 'coerce')\n",
    "\n",
    "mean_q4_balance = df['q4_balance'].mean()\n",
    "df['q4_balance'] = df['q4_balance'].fillna(mean_q4_balance)\n",
    "df['q4_balance'] = pd.to_numeric(df['q4_balance'], errors = 'coerce')\n",
    "\n",
    "#df.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0        -792.0\n",
       "1          67.0\n",
       "2         -15.0\n",
       "3         182.0\n",
       "4        -783.0\n",
       "          ...  \n",
       "17995     -71.0\n",
       "17996     -44.0\n",
       "17997    -151.0\n",
       "17998     357.0\n",
       "17999    1113.0\n",
       "Name: dif_q1_q2, Length: 18000, dtype: float64"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# make a new column that gives the difference between q2 and q1\n",
    "df['dif_q1_q2']= df['q2_balance']-df['q1_balance']\n",
    "df['dif_q1_q2']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0         -93.0\n",
       "1         288.0\n",
       "2          97.0\n",
       "3        -700.0\n",
       "4        -617.0\n",
       "          ...  \n",
       "17995     327.0\n",
       "17996    -408.0\n",
       "17997     324.0\n",
       "17998     750.0\n",
       "17999   -1375.0\n",
       "Name: dif_q3_q4, Length: 18000, dtype: float64"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# make a new column that gives the difference between q2 and q1\n",
    "df['dif_q3_q4']= df['q4_balance']-df['q3_balance']\n",
    "df['dif_q3_q4']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['customer_number', 'offer_accepted', 'reward', 'mailer_type',\n",
       "       'income_level', 'bank_accounts_open', 'overdraft_protection',\n",
       "       'credit_rating', 'credit_cards_held', 'homes_owned', 'household_size',\n",
       "       'own_your_home', 'average_balance', 'q1_balance', 'q2_balance',\n",
       "       'q3_balance', 'q4_balance', 'dif_q1_q2', 'dif_q3_q4'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3    5712\n",
       "4    5543\n",
       "5    2853\n",
       "2    2813\n",
       "1     539\n",
       "6     538\n",
       "9       1\n",
       "8       1\n",
       "Name: household_size, dtype: int64"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['household_size'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Get the sweet summary stats on each column to understand the distribution better <a class=\"anchor\" id=\"summary\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "#statistical distribution\n",
    "#df.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Get a  representation of the numbers  <a class=\"anchor\" id=\"numbersrep\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# using pandas\n",
    "# hist = df.hist(figsize=(15,10),layout=(3,4), grid=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # check relationships with seaborn\n",
    "# sns.pairplot(df)\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<B>Only the household_size has a normal distribution.<B>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # check for multicollinearity \n",
    "# corr_matrix=df.corr(method='pearson')  # default\n",
    "# fig, ax = plt.subplots(figsize=(10, 8))\n",
    "# ax = sns.heatmap(corr_matrix, annot=True)\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The graph proves the absence of multicollinearity. \n",
    "\n",
    "\"The linear model makes the following assumptions\n",
    "- Require a linear relationship between the dependent and independent variables.\n",
    "- The error terms (residuals) need to be normally distributed. \n",
    "- If not weights can be introduced to penalize some observations in the error metric.\n",
    "Independent variables are normally distributed.\n",
    "If not (apply transformation Box-Cox)\n",
    "- Absence of multicollinearity.\n",
    "- Independent observations (not repeated)\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "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>offer_accepted</th>\n",
       "      <th>reward</th>\n",
       "      <th>mailer_type</th>\n",
       "      <th>income_level</th>\n",
       "      <th>bank_accounts_open</th>\n",
       "      <th>credit_rating</th>\n",
       "      <th>credit_cards_held</th>\n",
       "      <th>homes_owned</th>\n",
       "      <th>household_size</th>\n",
       "      <th>own_your_home</th>\n",
       "      <th>dif_q1_q2</th>\n",
       "      <th>dif_q3_q4</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>1</td>\n",
       "      <td>High</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>No</td>\n",
       "      <td>-792.0</td>\n",
       "      <td>-93.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>Medium</td>\n",
       "      <td>1</td>\n",
       "      <td>Medium</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>Yes</td>\n",
       "      <td>67.0</td>\n",
       "      <td>288.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>High</td>\n",
       "      <td>2</td>\n",
       "      <td>Medium</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>Yes</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>97.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>Medium</td>\n",
       "      <td>2</td>\n",
       "      <td>High</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>No</td>\n",
       "      <td>182.0</td>\n",
       "      <td>-700.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>Medium</td>\n",
       "      <td>1</td>\n",
       "      <td>Medium</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>Yes</td>\n",
       "      <td>-783.0</td>\n",
       "      <td>-617.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>Medium</td>\n",
       "      <td>1</td>\n",
       "      <td>High</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>No</td>\n",
       "      <td>-482.0</td>\n",
       "      <td>-253.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>Medium</td>\n",
       "      <td>1</td>\n",
       "      <td>Medium</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>No</td>\n",
       "      <td>-280.0</td>\n",
       "      <td>-215.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>No</td>\n",
       "      <td>Cash Back</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Low</td>\n",
       "      <td>1</td>\n",
       "      <td>Medium</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>Yes</td>\n",
       "      <td>259.0</td>\n",
       "      <td>21.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Medium</td>\n",
       "      <td>1</td>\n",
       "      <td>Low</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>Yes</td>\n",
       "      <td>-3.0</td>\n",
       "      <td>471.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>2</td>\n",
       "      <td>Low</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>Yes</td>\n",
       "      <td>-360.0</td>\n",
       "      <td>-399.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  offer_accepted     reward mailer_type income_level  bank_accounts_open  \\\n",
       "0             No  Air Miles      Letter         High                   1   \n",
       "1             No  Air Miles      Letter       Medium                   1   \n",
       "2             No  Air Miles    Postcard         High                   2   \n",
       "3             No  Air Miles      Letter       Medium                   2   \n",
       "4             No  Air Miles      Letter       Medium                   1   \n",
       "5             No  Air Miles      Letter       Medium                   1   \n",
       "6             No  Air Miles      Letter       Medium                   1   \n",
       "7             No  Cash Back    Postcard          Low                   1   \n",
       "8             No  Air Miles    Postcard       Medium                   1   \n",
       "9             No  Air Miles      Letter         High                   2   \n",
       "\n",
       "  credit_rating  credit_cards_held  homes_owned  household_size own_your_home  \\\n",
       "0          High                  2            1               4            No   \n",
       "1        Medium                  2            2               5           Yes   \n",
       "2        Medium                  2            1               2           Yes   \n",
       "3          High                  1            1               4            No   \n",
       "4        Medium                  2            1               6           Yes   \n",
       "5          High                  3            1               4            No   \n",
       "6        Medium                  2            1               3            No   \n",
       "7        Medium                  4            1               4           Yes   \n",
       "8           Low                  2            1               4           Yes   \n",
       "9           Low                  3            2               4           Yes   \n",
       "\n",
       "   dif_q1_q2  dif_q3_q4  \n",
       "0     -792.0      -93.0  \n",
       "1       67.0      288.0  \n",
       "2      -15.0       97.0  \n",
       "3      182.0     -700.0  \n",
       "4     -783.0     -617.0  \n",
       "5     -482.0     -253.0  \n",
       "6     -280.0     -215.0  \n",
       "7      259.0       21.0  \n",
       "8       -3.0      471.0  \n",
       "9     -360.0     -399.0  "
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check for more claning\n",
    "# df.head(20)\n",
    "# df.tail(20)\n",
    "# drop customer_number\n",
    "df=df.drop(['customer_number','average_balance','q1_balance', 'q2_balance',\n",
    "       'q3_balance', 'q4_balance','overdraft_protection',], axis=1)\n",
    "df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It seems that there is no need to make more cleaning. Neither for Bucketing features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['offer_accepted', 'reward', 'mailer_type', 'income_level',\n",
       "       'bank_accounts_open', 'credit_rating', 'credit_cards_held',\n",
       "       'homes_owned', 'household_size', 'own_your_home', 'dif_q1_q2',\n",
       "       'dif_q3_q4'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###  New dataframe for binning <a class=\"anchor\" id=\"eda\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.groupby(['offer_accepted', 'reward', 'mailer_type', 'income_level','credit_rating','household_size', 'credit_cards_held',\n",
    "       'homes_owned', 'own_your_home', 'dif_q1_q2',\n",
    "       'dif_q3_q4'])['bank_accounts_open'].sum().reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# common practise to keep the orginal data\n",
    "df['bank_accounts_open'] = pd.qcut(df['bank_accounts_open'],duplicates='drop', q=6)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(0.999, 2.0], (2.0, 3.0]]\n",
       "Categories (2, interval[float64]): [(0.999, 2.0] < (2.0, 3.0]]"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['bank_accounts_open'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "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>offer_accepted</th>\n",
       "      <th>reward</th>\n",
       "      <th>mailer_type</th>\n",
       "      <th>income_level</th>\n",
       "      <th>credit_rating</th>\n",
       "      <th>household_size</th>\n",
       "      <th>credit_cards_held</th>\n",
       "      <th>homes_owned</th>\n",
       "      <th>own_your_home</th>\n",
       "      <th>dif_q1_q2</th>\n",
       "      <th>dif_q3_q4</th>\n",
       "      <th>bank_accounts_open</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>High</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>Yes</td>\n",
       "      <td>-185.0</td>\n",
       "      <td>424.0</td>\n",
       "      <td>(0.999, 2.0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>High</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>Yes</td>\n",
       "      <td>-467.0</td>\n",
       "      <td>-399.0</td>\n",
       "      <td>(0.999, 2.0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>High</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>No</td>\n",
       "      <td>-656.0</td>\n",
       "      <td>-570.0</td>\n",
       "      <td>(0.999, 2.0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>High</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>No</td>\n",
       "      <td>95.0</td>\n",
       "      <td>-121.0</td>\n",
       "      <td>(0.999, 2.0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>High</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>Yes</td>\n",
       "      <td>-561.0</td>\n",
       "      <td>-875.0</td>\n",
       "      <td>(0.999, 2.0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17971</th>\n",
       "      <td>Yes</td>\n",
       "      <td>Points</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Medium</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>Yes</td>\n",
       "      <td>910.0</td>\n",
       "      <td>-580.0</td>\n",
       "      <td>(0.999, 2.0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17972</th>\n",
       "      <td>Yes</td>\n",
       "      <td>Points</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Medium</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>Yes</td>\n",
       "      <td>-379.0</td>\n",
       "      <td>-709.0</td>\n",
       "      <td>(0.999, 2.0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17973</th>\n",
       "      <td>Yes</td>\n",
       "      <td>Points</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Medium</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>Yes</td>\n",
       "      <td>603.0</td>\n",
       "      <td>-231.0</td>\n",
       "      <td>(0.999, 2.0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17974</th>\n",
       "      <td>Yes</td>\n",
       "      <td>Points</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Medium</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>Yes</td>\n",
       "      <td>1031.0</td>\n",
       "      <td>139.0</td>\n",
       "      <td>(0.999, 2.0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17975</th>\n",
       "      <td>Yes</td>\n",
       "      <td>Points</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Medium</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>Yes</td>\n",
       "      <td>295.0</td>\n",
       "      <td>515.0</td>\n",
       "      <td>(0.999, 2.0]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>17976 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      offer_accepted     reward mailer_type income_level credit_rating  \\\n",
       "0                 No  Air Miles      Letter         High          High   \n",
       "1                 No  Air Miles      Letter         High          High   \n",
       "2                 No  Air Miles      Letter         High          High   \n",
       "3                 No  Air Miles      Letter         High          High   \n",
       "4                 No  Air Miles      Letter         High          High   \n",
       "...              ...        ...         ...          ...           ...   \n",
       "17971            Yes     Points    Postcard       Medium        Medium   \n",
       "17972            Yes     Points    Postcard       Medium        Medium   \n",
       "17973            Yes     Points    Postcard       Medium        Medium   \n",
       "17974            Yes     Points    Postcard       Medium        Medium   \n",
       "17975            Yes     Points    Postcard       Medium        Medium   \n",
       "\n",
       "       household_size  credit_cards_held  homes_owned own_your_home  \\\n",
       "0                   1                  1            1           Yes   \n",
       "1                   1                  1            2           Yes   \n",
       "2                   1                  2            1            No   \n",
       "3                   1                  2            1            No   \n",
       "4                   1                  2            1           Yes   \n",
       "...               ...                ...          ...           ...   \n",
       "17971               4                  2            2           Yes   \n",
       "17972               4                  3            1           Yes   \n",
       "17973               4                  3            1           Yes   \n",
       "17974               4                  3            1           Yes   \n",
       "17975               6                  3            1           Yes   \n",
       "\n",
       "       dif_q1_q2  dif_q3_q4 bank_accounts_open  \n",
       "0         -185.0      424.0       (0.999, 2.0]  \n",
       "1         -467.0     -399.0       (0.999, 2.0]  \n",
       "2         -656.0     -570.0       (0.999, 2.0]  \n",
       "3           95.0     -121.0       (0.999, 2.0]  \n",
       "4         -561.0     -875.0       (0.999, 2.0]  \n",
       "...          ...        ...                ...  \n",
       "17971      910.0     -580.0       (0.999, 2.0]  \n",
       "17972     -379.0     -709.0       (0.999, 2.0]  \n",
       "17973      603.0     -231.0       (0.999, 2.0]  \n",
       "17974     1031.0      139.0       (0.999, 2.0]  \n",
       "17975      295.0      515.0       (0.999, 2.0]  \n",
       "\n",
       "[17976 rows x 12 columns]"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Outliers <a class=\"anchor\" id=\"outliers\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Box Plots <a class=\"anchor\" id=\"boxplots\"></a>\n",
    "A boxplot is a standardized way of displaying the distribution of data based on a five number summary (“minimum”, first quartile (Q1), median, third quartile (Q3), and “maximum”). It can tell you about your outliers and what their values are. It can also tell you if your data is symmetrical, how tightly your data is grouped, and if and how your data is skewed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Q1 of the column  dif_q1_q2 : -300.0\n",
      "Q3 of the column  dif_q1_q2 : 411.0\n",
      "IRQ of the column  dif_q1_q2 : 711.0\n",
      "lower_whisker:  -1366.5 , upper whisker:  1477.5\n",
      "Q1 of the column  dif_q3_q4 : -558.0\n",
      "Q3 of the column  dif_q3_q4 : 105.0\n",
      "IRQ of the column  dif_q3_q4 : 663.0\n",
      "lower_whisker:  -1552.5 , upper whisker:  1099.5\n"
     ]
    }
   ],
   "source": [
    "# find quartiles and clean up outliers\n",
    "cols_outliers = ['dif_q1_q2',\n",
    "       'dif_q3_q4']\n",
    "\n",
    "for col in cols_outliers:\n",
    "    Q1 = df[col].quantile(0.25)\n",
    "    Q3 = df[col].quantile(0.75)\n",
    "    IQR = Q3 - Q1\n",
    "    print('Q1 of the column ', col,':', Q1)\n",
    "    print('Q3 of the column ', col,':', Q3)\n",
    "    print('IRQ of the column ', col,':', IQR)\n",
    "    Lower_Whisker = Q1 - 1.5 * IQR\n",
    "    Upper_Whisker = Q3 + 1.5 * IQR\n",
    "    print('lower_whisker: ', Lower_Whisker, ',' , 'upper whisker: ', Upper_Whisker)\n",
    "    #  Outlier Treatment \n",
    "    df = df[df[col]< Upper_Whisker]\n",
    "    df = df[df[col]> Lower_Whisker]\n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#check outliers\n",
    "df.boxplot(grid=False, figsize= (15,8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# # check for multicollinearity \n",
    "corr_matrix=df.corr(method='pearson')  # default\n",
    "fig, ax = plt.subplots(figsize=(10, 8))\n",
    "ax = sns.heatmap(corr_matrix, annot=True)\n",
    "plt.show()\n",
    "# q1, q2 ,q3 have collinearity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "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>offer_accepted</th>\n",
       "      <th>reward</th>\n",
       "      <th>mailer_type</th>\n",
       "      <th>income_level</th>\n",
       "      <th>credit_rating</th>\n",
       "      <th>own_your_home</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>High</td>\n",
       "      <td>Yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>High</td>\n",
       "      <td>Yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>High</td>\n",
       "      <td>No</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>High</td>\n",
       "      <td>No</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>No</td>\n",
       "      <td>Air Miles</td>\n",
       "      <td>Letter</td>\n",
       "      <td>High</td>\n",
       "      <td>High</td>\n",
       "      <td>Yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17971</th>\n",
       "      <td>Yes</td>\n",
       "      <td>Points</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17972</th>\n",
       "      <td>Yes</td>\n",
       "      <td>Points</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17973</th>\n",
       "      <td>Yes</td>\n",
       "      <td>Points</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17974</th>\n",
       "      <td>Yes</td>\n",
       "      <td>Points</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17975</th>\n",
       "      <td>Yes</td>\n",
       "      <td>Points</td>\n",
       "      <td>Postcard</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Medium</td>\n",
       "      <td>Yes</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>17519 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      offer_accepted     reward mailer_type income_level credit_rating  \\\n",
       "0                 No  Air Miles      Letter         High          High   \n",
       "1                 No  Air Miles      Letter         High          High   \n",
       "2                 No  Air Miles      Letter         High          High   \n",
       "3                 No  Air Miles      Letter         High          High   \n",
       "4                 No  Air Miles      Letter         High          High   \n",
       "...              ...        ...         ...          ...           ...   \n",
       "17971            Yes     Points    Postcard       Medium        Medium   \n",
       "17972            Yes     Points    Postcard       Medium        Medium   \n",
       "17973            Yes     Points    Postcard       Medium        Medium   \n",
       "17974            Yes     Points    Postcard       Medium        Medium   \n",
       "17975            Yes     Points    Postcard       Medium        Medium   \n",
       "\n",
       "      own_your_home  \n",
       "0               Yes  \n",
       "1               Yes  \n",
       "2                No  \n",
       "3                No  \n",
       "4               Yes  \n",
       "...             ...  \n",
       "17971           Yes  \n",
       "17972           Yes  \n",
       "17973           Yes  \n",
       "17974           Yes  \n",
       "17975           Yes  \n",
       "\n",
       "[17519 rows x 6 columns]"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# label / encode any categorical columns \n",
    "\n",
    "cat = df.select_dtypes(include = np.object)\n",
    "cat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # checking all the categorical columns\n",
    "# cols_cat = ['offer_accepted', 'reward', 'mailer_type', 'income_level', 'overdraft_protection', 'credit_rating', 'own_your_home']\n",
    "# for col in cols_cat:\n",
    "#     print(\"Frequency analysis of column: \",col)\n",
    "#     my_data = df[col].value_counts().reset_index()\n",
    "#     my_data.drop(my_data.index[0])\n",
    "#     ax = sns.barplot(x=col, y=\"index\", data = my_data).set_title(col.upper())\n",
    "#     plt.figure()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "#a =df['average_balance'].value_counts().reset_index()\n",
    "#a.drop(a.index[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "# locate the target column in the last position\n",
    "cols = list(df.columns)\n",
    "cols = cols[1:]+[cols[0]]\n",
    "df = df[cols]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert balances to float\n",
    "\n",
    "#df['q2_balance'] = pd.to_numeric(df.q2_balance, errors='coerce')\n",
    "#df['q3_balance'] = pd.to_numeric(df.q3_balance, errors='coerce')\n",
    "#df['q4_balance'] = pd.to_numeric(df.q4_balance, errors='coerce')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df.dtypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Density Plots\n",
    "\n",
    "A density plot is a representation of the distribution of a numeric variable. It uses a kernel density estimate to show the probability density function of the variable. It is a smoothed version of the histogram and is used in the same concept.\n",
    "\n",
    "Kernel Density Estimation(KDE): Kernel density estimation is a really useful statistical tool with an intimidating name. Often shortened to KDE, it’s a technique that let’s you create a smooth curve given a set of data. Kernel Density Estimation (KDE) is a way to estimate the probability density function of a continuous random variable. It is used for non-parametric analysis."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "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>household_size</th>\n",
       "      <th>credit_cards_held</th>\n",
       "      <th>homes_owned</th>\n",
       "      <th>dif_q1_q2</th>\n",
       "      <th>dif_q3_q4</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>-185.0</td>\n",
       "      <td>424.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>-467.0</td>\n",
       "      <td>-399.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>-656.0</td>\n",
       "      <td>-570.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>95.0</td>\n",
       "      <td>-121.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>-561.0</td>\n",
       "      <td>-875.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17971</th>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>910.0</td>\n",
       "      <td>-580.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17972</th>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>-379.0</td>\n",
       "      <td>-709.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17973</th>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>603.0</td>\n",
       "      <td>-231.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17974</th>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1031.0</td>\n",
       "      <td>139.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17975</th>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>295.0</td>\n",
       "      <td>515.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>17519 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       household_size  credit_cards_held  homes_owned  dif_q1_q2  dif_q3_q4\n",
       "0                   1                  1            1     -185.0      424.0\n",
       "1                   1                  1            2     -467.0     -399.0\n",
       "2                   1                  2            1     -656.0     -570.0\n",
       "3                   1                  2            1       95.0     -121.0\n",
       "4                   1                  2            1     -561.0     -875.0\n",
       "...               ...                ...          ...        ...        ...\n",
       "17971               4                  2            2      910.0     -580.0\n",
       "17972               4                  3            1     -379.0     -709.0\n",
       "17973               4                  3            1      603.0     -231.0\n",
       "17974               4                  3            1     1031.0      139.0\n",
       "17975               6                  3            1      295.0      515.0\n",
       "\n",
       "[17519 rows x 5 columns]"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "numerics = ['int64', 'float64']\n",
    "df_num = df.select_dtypes(include=numerics)\n",
    "df_num\n",
    "# for col in df_num.columns:\n",
    "#     ax = sns.kdeplot(data= df_num, x= col, shade= True)\n",
    "#     plt.figure()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for col in df_num.columns:\n",
    "#     plt.figure(figsize=(10,6)) \n",
    "#     sns.kdeplot(df[col][df.offer_accepted  == 'Yes'], shade=True)\n",
    "#     sns.kdeplot(df[col][df.offer_accepted  == 'No'], shade=True)\n",
    "#     plt.legend(['offer_accepted ', 'offer_rejected'])\n",
    "#     plt.title(col)\n",
    "#     plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "# check outliers\n",
    "#df.boxplot(grid=False, figsize= (15,8))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note: the quartiles didn't removed all the outliers for average_balance. It could be improved."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<AxesSubplot:title={'center':'household_size'}>,\n",
       "        <AxesSubplot:title={'center':'credit_cards_held'}>,\n",
       "        <AxesSubplot:title={'center':'homes_owned'}>,\n",
       "        <AxesSubplot:title={'center':'dif_q1_q2'}>],\n",
       "       [<AxesSubplot:title={'center':'dif_q3_q4'}>, <AxesSubplot:>,\n",
       "        <AxesSubplot:>, <AxesSubplot:>],\n",
       "       [<AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 12 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.hist(figsize=(15,10),layout=(3,4), grid=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pre-processing <a class=\"anchor\" id=\"preprocessing\"></a>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "#cat.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "#cat.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "# reset index\n",
    "df = df.reset_index(drop = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 17519 entries, 0 to 17518\n",
      "Data columns (total 10 columns):\n",
      " #   Column              Non-Null Count  Dtype   \n",
      "---  ------              --------------  -----   \n",
      " 0   reward              17519 non-null  object  \n",
      " 1   mailer_type         17519 non-null  object  \n",
      " 2   income_level        17519 non-null  object  \n",
      " 3   credit_rating       17519 non-null  object  \n",
      " 4   household_size      17519 non-null  int64   \n",
      " 5   credit_cards_held   17519 non-null  int64   \n",
      " 6   homes_owned         17519 non-null  int64   \n",
      " 7   own_your_home       17519 non-null  object  \n",
      " 8   bank_accounts_open  17519 non-null  category\n",
      " 9   offer_accepted      17519 non-null  object  \n",
      "dtypes: category(1), int64(3), object(6)\n",
      "memory usage: 1.2+ MB\n"
     ]
    }
   ],
   "source": [
    "# categorical = pd.get_dummies(cat, columns =['reward', 'mailer_type', 'income_level',\n",
    "#        'overdraft_protection', 'credit_rating', 'own_your_home',\n",
    "#        'average_balance', 'offer_accepted'],drop_first=True)\n",
    "\n",
    "cat = df.select_dtypes(exclude = ['float64'])\n",
    "cat.info()\n",
    "\n",
    "# categorical.head()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['reward',\n",
       " 'mailer_type',\n",
       " 'income_level',\n",
       " 'credit_rating',\n",
       " 'household_size',\n",
       " 'credit_cards_held',\n",
       " 'homes_owned',\n",
       " 'own_your_home',\n",
       " 'bank_accounts_open',\n",
       " 'offer_accepted']"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# create a list of the column names\n",
    "cat_col=cat.columns.to_list()\n",
    "cat_col"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "categorical = pd.get_dummies(cat, columns =['reward',\n",
    " 'mailer_type',\n",
    " 'income_level',\n",
    " 'bank_accounts_open',\n",
    " 'credit_rating',\n",
    " 'credit_cards_held',\n",
    " 'homes_owned',\n",
    " 'own_your_home',\n",
    " 'household_size',\n",
    " 'offer_accepted'],drop_first=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "#categorical.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "#categorical['offer_accepted_Yes'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "# percentage\n",
    "#pct_offer_accepted = 1023/18000\n",
    "#pct_offer_accepted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "# numerical scaling \n",
    "from sklearn.preprocessing import Normalizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "# df_num are the numerical\n",
    "\n",
    "X = df_num.drop(['household_size', 'credit_cards_held', 'homes_owned'], axis=1)\n",
    "transformer = Normalizer().fit(X)\n",
    "X_normalised= transformer.transform(X)\n",
    "X_normalised = pd.DataFrame(X_normalised)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "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>0</th>\n",
       "      <th>1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.399911</td>\n",
       "      <td>0.916554</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.760290</td>\n",
       "      <td>-0.649584</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.754853</td>\n",
       "      <td>-0.655894</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.617534</td>\n",
       "      <td>-0.786544</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.539736</td>\n",
       "      <td>-0.841834</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17514</th>\n",
       "      <td>0.843280</td>\n",
       "      <td>-0.537475</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17515</th>\n",
       "      <td>-0.471427</td>\n",
       "      <td>-0.881905</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17516</th>\n",
       "      <td>0.933824</td>\n",
       "      <td>-0.357733</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17517</th>\n",
       "      <td>0.991034</td>\n",
       "      <td>0.133612</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17518</th>\n",
       "      <td>0.497046</td>\n",
       "      <td>0.867724</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>17519 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              0         1\n",
       "0     -0.399911  0.916554\n",
       "1     -0.760290 -0.649584\n",
       "2     -0.754853 -0.655894\n",
       "3      0.617534 -0.786544\n",
       "4     -0.539736 -0.841834\n",
       "...         ...       ...\n",
       "17514  0.843280 -0.537475\n",
       "17515 -0.471427 -0.881905\n",
       "17516  0.933824 -0.357733\n",
       "17517  0.991034  0.133612\n",
       "17518  0.497046  0.867724\n",
       "\n",
       "[17519 rows x 2 columns]"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_normalised\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Split off the dependant variable (label) <a class=\"anchor\" id=\"split\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "#sort out the categoricals \n",
    "y= categorical['offer_accepted_Yes']\n",
    "categorical=categorical.drop(['offer_accepted_Yes'], axis=1)\n",
    "#categorical"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "#y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "#categorical"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "# bring the numerical data (scaled)in with the x cat\n",
    "X = np.concatenate((X_normalised, categorical),axis=1)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### TT split, get model <a class=\"anchor\" id=\"ttsplit\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_train, X_test, y_train, y_test =train_test_split(X, y, test_size= 0.4, random_state=40)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Apply model and train model <a class=\"anchor\" id=\"applymodel\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "classification = LogisticRegression(random_state=0, solver='lbfgs',max_iter=400,\n",
    "                  multi_class='ovr').fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Evaluate accuracy and test <a class=\"anchor\" id=\"evaluate\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score \n",
    "predictions = classification.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 0, ..., 0, 0, 0], dtype=uint8)"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9434931506849316"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(y_test, predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "# next steps would be to visualise the accuracy of the predictions in some ways \n",
    "\n",
    "# can i improve my model ?? "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualising accuracy - ROC / AUC  <a class=\"anchor\" id=\"rocauc\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.7495871448911376\n"
     ]
    }
   ],
   "source": [
    "import sklearn.metrics as metrics \n",
    "probs = classification.predict_proba(X_test)\n",
    "preds =probs[:,1]\n",
    "fpr, tpr, threshold = metrics.roc_curve(y_test, preds)\n",
    "roc_auc = metrics.auc(fpr, tpr)\n",
    "print(roc_auc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "#definitions \n",
    "# tpr = true positive rate \n",
    "# fpr = false positive rate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "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": [
    "#method to plot AUC visually \n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr, tpr, 'b', label = 'AUC = %0.2f' % roc_auc)\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "# confusion matrix \n",
    "\n",
    "from sklearn.metrics import confusion_matrix \n",
    "from sklearn.metrics import plot_confusion_matrix\n",
    "\n",
    "#accuracy_score(y_test,predictions)\n",
    "#confusion_matrix(y_test,predictions)\n",
    "#plot_confusion_matrix(classification, X_test, y_test)\n",
    "#plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data imbalanced -  SMOTE (oversample the minority class of the target label) <a class=\"anchor\" id=\"smote\"></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "from imblearn.over_sampling import SMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "smote = SMOTE()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    16522\n",
       "0    16522\n",
       "Name: offer_accepted_Yes, dtype: int64"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_sm, y_sm = smote.fit_resample(X, y)\n",
    "y_sm.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "#re do train test split It2\n",
    "\n",
    "X_train, X_test, y_train, y_test =train_test_split(X_sm, y_sm, test_size= 0.4, random_state=40)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "# apply model It2 \n",
    "classification = LogisticRegression(random_state=0, solver='lbfgs',max_iter=400,\n",
    "                  multi_class='ovr').fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7193221364805568"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#accuracy assessment It2\n",
    "predictions = classification.predict(X_test)\n",
    "accuracy_score(y_test, predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "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": [
    "#method to plot AUC visually \n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr, tpr, 'b', label = 'AUC = %0.2f' % roc_auc)\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Conf matrix It2 \n",
    "confusion_matrix(y_test,predictions)\n",
    "plot_confusion_matrix(classification, X_test, y_test)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### conclusion on imbalance:\n",
    "\n",
    "- The SMOTE technique solved the imbalanced but the accuracy got too low (71%), but the TOMEK doesn't.\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### References\n",
    "- lectures\n",
    "- women who code: munibootcamp of data science\n",
    "- https://machinelearningmastery.com/confusion-matrix-machine-learning/\n",
    "- https://medium.com/analytics-vidhya/outlier-treatment-9bbe87384d02\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.8.5"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": false,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {
    "height": "789px",
    "left": "108px",
    "top": "110px",
    "width": "207.594px"
   },
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}