Data Imputation & Cleaning - Phase_1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Modules :\n",
    "\n",
    "from pyforest import *\n",
    "from tqdm import tqdm\n",
    "import warnings\n",
    "from IPython.display import Image\n",
    "import re\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.svm import SVC\n",
    "from prettytable import PrettyTable\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![title](Merchant_Banking.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# AmExpert 2019 – Machine Learning Hackathon\n",
    "\n",
    "\n",
    "## Introduction\n",
    "American Express and Analytics Vidhya presents “AmExpert 2019 – Machine Learning Hackathon”. An amazing opportunity to showcase your analytical abilities and talent.\n",
    "\n",
    "Get a taste of the kind of challenges we face here at American Express on day to day basis.\n",
    "\n",
    "Exciting prizes up for grabs! Best performers also get a chance to get interviewed by American Express for analytics roles!\n",
    "\n",
    "Don’t hold on to the buzzing ideas in your mind. Just wear your thinking hats and display the spectrum of your creativity!\n",
    "\n",
    "\n",
    "# Problem Statement\n",
    "\n",
    "## Predicting Coupon Redemption\n",
    "\n",
    "XYZ Credit Card company regularly helps it’s merchants understand their data better and take key business decisions accurately by providing machine learning and analytics consulting. ABC is an established Brick & Mortar retailer that frequently conducts marketing campaigns for its diverse product range. As a merchant of XYZ, they have sought XYZ to assist them in their discount marketing process using the power of machine learning. Can you wear the AmExpert hat and help out ABC?\n",
    "\n",
    " \n",
    "Discount marketing and coupon usage are very widely used promotional techniques to attract new customers and to retain & reinforce loyalty of existing customers. The measurement of a consumer’s propensity towards coupon usage and the prediction of the redemption behaviour are crucial parameters in assessing the effectiveness of a marketing campaign.\n",
    "\n",
    " \n",
    "ABC’s promotions are shared across various channels including email, notifications, etc. A number of these campaigns include coupon discounts that are offered for a specific product/range of products. The retailer would like the ability to predict whether customers redeem the coupons received across channels, which will enable the retailer’s marketing team to accurately design coupon construct, and develop more precise and targeted marketing strategies.\n",
    "\n",
    " \n",
    "The data available in this problem contains the following information, including the details of a sample of campaigns and coupons used in previous campaigns -\n",
    "\n",
    "**User Demographic Details**\n",
    "\n",
    "**Campaign and coupon Details**\n",
    "\n",
    "**Product details**\n",
    "\n",
    "**Previous transactions**\n",
    "\n",
    "\n",
    "Based on previous transaction & performance data from the last 18 campaigns, predict the probability for the next 10 campaigns in the test set for each coupon and customer combination, whether the customer will redeem the coupon or not?\n",
    "\n",
    " \n",
    "\n",
    "### Dataset Description\n",
    "\n",
    "Here is the schema for the different data tables available. The detailed data dictionary is provided next.\n",
    "\n",
    "![title](amex19.png)\n",
    "\n",
    "You are provided with the following files in train.zip:\n",
    "\n",
    "**train.csv:** Train data containing the coupons offered to the given customers under the 18 campaigns\n",
    "\n",
    "|Variable |Definition |\n",
    "|----|----|\n",
    "|id\t|Unique id for coupon customer impression|\n",
    "|campaign_id\t|Unique id for a discount campaign|\n",
    "|coupon_id\t|Unique id for a discount coupon|\n",
    "|customer_id\t|Unique id for a customer|\n",
    "|redemption_status|\t(target) (0 - Coupon not redeemed, 1 - Coupon redeemed) |\n",
    "\n",
    "\n",
    "**campaign_data.csv:** Campaign information for each of the 28 campaigns\n",
    "\n",
    "|Variable|\tDefinition|\n",
    "|----|----|\n",
    "|campaign_id\t|Unique id for a discount campaign|\n",
    "|campaign_type\t|Anonymised Campaign Type (X/Y)|\n",
    "|start_date\t|Campaign Start Date|\n",
    "|end_date\t|Campaign End Date |\n",
    "\n",
    "\n",
    "**coupon_item_mapping.csv:** Mapping of coupon and items valid for discount under that coupon\n",
    "\n",
    "|Variable|\tDefinition|\n",
    "|----|----|\n",
    "|coupon_id\t|Unique id for a discount coupon (no order)|\n",
    "|item_id\t|Unique id for items for which given coupon is valid (no order) |\n",
    "\n",
    "\n",
    "**customer_demographics.csv:** Customer demographic information for some customers\n",
    "\n",
    "|Variable\t|Definition|\n",
    "|----|----|\n",
    "|customer_id\t|Unique id for a customer|\n",
    "|age_range\t|Age range of customer family in years|\n",
    "|marital_status|\tMarried/Single|\n",
    "|rented|\t0 - not rented accommodation, 1 - rented accommodation|\n",
    "|family_size|\tNumber of family members|\n",
    "|no_of_children\t|Number of children in the family|\n",
    "|income_bracket|\tLabel Encoded Income Bracket (Higher income corresponds to higher number) |\n",
    "\n",
    "\n",
    "**customer_transaction_data.csv:** Transaction data for all customers for duration of campaigns in the train data\n",
    "\n",
    "|Variable\t|Definition|\n",
    "|----|----|\n",
    "|date\t|Date of Transaction|\n",
    "|customer_id|Unique id for a customer|\n",
    "|item_id|\tUnique id for item|\n",
    "|quantity\t|quantity of item bought|\n",
    "|selling_price\t|Sales value of the transaction|\n",
    "|other_discount|\tDiscount from other sources such as manufacturer coupon/loyalty card|\n",
    "|coupon_discount|\tDiscount availed from retailer coupon |\n",
    "\n",
    "\n",
    "**item_data.csv:** Item information for each item sold by the retailer\n",
    "\n",
    "|Variable|\tDefinition|\n",
    "|----|----|\n",
    "|item_id|\tUnique id for itemv\n",
    "|brand\t|Unique id for item brand|\n",
    "|brand_type\t|Brand Type (local/Established)|\n",
    "|category\t|Item Category |\n",
    "\n",
    "\n",
    "**test.csv:** Contains the coupon customer combination for which redemption status is to be predicted\n",
    "\n",
    "|Variable|\tDefinition|\n",
    "|----|----|\n",
    "|id\t|Unique id for coupon customer impression|\n",
    "|campaign_id\t|Unique id for a discount campaign|\n",
    "|coupon_id\t|Unique id for a discount coupon|\n",
    "|customer_id\t|Unique id for a customer |\n",
    "\n",
    "*Campaign, coupon and customer data for test set is also contained in train.zip *\n",
    "\n",
    "**sample_submission.csv:** This file contains the format in which you have to submit your predictions.\n",
    "\n",
    "To summarise the entire process:\n",
    "\n",
    "* Customers receive coupons under various campaigns and may choose to redeem it.\n",
    "* They can redeem the given coupon for any valid product for that coupon as per coupon item mapping within the duration between campaign start date and end date\n",
    "* Next, the customer will redeem the coupon for an item at the retailer store and that will reflect in the transaction table in the column coupon_discount.\n",
    " \n",
    "### Evaluation Metric\n",
    "Submissions are evaluated on area under the ROC curve between the predicted probability and the observed target.\n",
    " \n",
    "\n",
    "### Public and Private Split\n",
    "* Test data is further randomly divided into Public (40%) and Private data (60%)\n",
    "* Your initial responses will be checked and scored on the Public data.\n",
    "* The final rankings would be based on your private score which will be published once the competition is over."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Importing Datasets :\n",
    "\n",
    "campaign_data = pd.read_csv(r'D:\\MOOCs & Material\\Data Science & ML\\Data Modelling - Datasets\\American Express (AmExpert) - Analytics vidhya\\Datasets\\campaign_data.csv')\n",
    "coupon_item_mapping = pd.read_csv(r'D:\\MOOCs & Material\\Data Science & ML\\Data Modelling - Datasets\\American Express (AmExpert) - Analytics vidhya\\Datasets\\coupon_item_mapping.csv')\n",
    "customer_demographs = pd.read_csv(r'D:\\MOOCs & Material\\Data Science & ML\\Data Modelling - Datasets\\American Express (AmExpert) - Analytics vidhya\\Datasets\\customer_demographics.csv')\n",
    "customer_transaction_data = pd.read_csv(r'D:\\MOOCs & Material\\Data Science & ML\\Data Modelling - Datasets\\American Express (AmExpert) - Analytics vidhya\\Datasets\\customer_transaction_data.csv')\n",
    "item_data = pd.read_csv(r'D:\\MOOCs & Material\\Data Science & ML\\Data Modelling - Datasets\\American Express (AmExpert) - Analytics vidhya\\Datasets\\item_data.csv')\n",
    "train = pd.read_csv(r'D:\\MOOCs & Material\\Data Science & ML\\Data Modelling - Datasets\\American Express (AmExpert) - Analytics vidhya\\Datasets\\train.csv')\n",
    "test = pd.read_csv(r'D:\\MOOCs & Material\\Data Science & ML\\Data Modelling - Datasets\\American Express (AmExpert) - Analytics vidhya\\Datasets\\test.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2>OBJECTIVE :</h2>\n",
    "\n",
    "Based on previous transaction & performance data from the last **18** campaigns, predict the probability for the next **10** campaigns in the test set for each coupon and customer combination, whether the customer will redeem the coupon or not?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2>Performing EDA for </h2>\n",
    "\n",
    "+ Campaign_Data\n",
    "+ Coupon_Item_Mapping\n",
    "+ Item_data\n",
    "+ Customer_Transaction_Data "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### *Campaign Data :* "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>campaign_id</th>\n",
       "      <th>campaign_type</th>\n",
       "      <th>start_date</th>\n",
       "      <th>end_date</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>24</td>\n",
       "      <td>Y</td>\n",
       "      <td>21/10/13</td>\n",
       "      <td>20/12/13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>25</td>\n",
       "      <td>Y</td>\n",
       "      <td>21/10/13</td>\n",
       "      <td>22/11/13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>20</td>\n",
       "      <td>Y</td>\n",
       "      <td>07/09/13</td>\n",
       "      <td>16/11/13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>23</td>\n",
       "      <td>Y</td>\n",
       "      <td>08/10/13</td>\n",
       "      <td>15/11/13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>21</td>\n",
       "      <td>Y</td>\n",
       "      <td>16/09/13</td>\n",
       "      <td>18/10/13</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   campaign_id campaign_type start_date  end_date\n",
       "0           24             Y   21/10/13  20/12/13\n",
       "1           25             Y   21/10/13  22/11/13\n",
       "2           20             Y   07/09/13  16/11/13\n",
       "3           23             Y   08/10/13  15/11/13\n",
       "4           21             Y   16/09/13  18/10/13"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "campaign_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 28 entries, 0 to 27\n",
      "Data columns (total 4 columns):\n",
      "campaign_id      28 non-null int64\n",
      "campaign_type    28 non-null object\n",
      "start_date       28 non-null object\n",
      "end_date         28 non-null object\n",
      "dtypes: int64(1), object(3)\n",
      "memory usage: 1.0+ KB\n"
     ]
    }
   ],
   "source": [
    "campaign_data.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "28"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(campaign_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "28"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Total no of unique campaign ids :\n",
    "\n",
    "campaign_data.campaign_id.nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "count_by_train :  18\n",
      "count_by_test :  10\n"
     ]
    }
   ],
   "source": [
    "# For train set, there are 18 campaign_ids and therefore we need to predict for 10 campigns of the test set :\n",
    "\n",
    "count_by_train = 0\n",
    "\n",
    "for each in train.campaign_id.unique():\n",
    "    if each in campaign_data.campaign_id.unique():\n",
    "        count_by_train+=1\n",
    "        \n",
    "print('count_by_train : ', count_by_train)\n",
    "\n",
    "count_by_test = 0\n",
    "\n",
    "for each in test.campaign_id.unique():\n",
    "    if each in campaign_data.campaign_id.unique():\n",
    "        count_by_test+=1\n",
    "        \n",
    "print('count_by_test : ', count_by_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Unique No of campaign_type : 2\n",
      "Types No of campaign_type : ['Y' 'X']\n"
     ]
    }
   ],
   "source": [
    "# FEATURE - Campaign_Type :\n",
    "\n",
    "print('Unique No of campaign_type :', campaign_data.campaign_type.nunique())\n",
    "print('Types No of campaign_type :', campaign_data.campaign_type.unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Converting to datetime for the 'start_date' and the 'end_date' of the campaigns :\n",
    "\n",
    "campaign_data.start_date = pd.to_datetime(campaign_data.start_date)\n",
    "campaign_data.end_date = pd.to_datetime(campaign_data.end_date)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "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>campaign_id</th>\n",
       "      <th>campaign_type</th>\n",
       "      <th>start_date</th>\n",
       "      <th>end_date</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>24</td>\n",
       "      <td>Y</td>\n",
       "      <td>2013-10-21</td>\n",
       "      <td>2013-12-20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>25</td>\n",
       "      <td>Y</td>\n",
       "      <td>2013-10-21</td>\n",
       "      <td>2013-11-22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>20</td>\n",
       "      <td>Y</td>\n",
       "      <td>2013-07-09</td>\n",
       "      <td>2013-11-16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>23</td>\n",
       "      <td>Y</td>\n",
       "      <td>2013-08-10</td>\n",
       "      <td>2013-11-15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>21</td>\n",
       "      <td>Y</td>\n",
       "      <td>2013-09-16</td>\n",
       "      <td>2013-10-18</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   campaign_id campaign_type start_date   end_date\n",
       "0           24             Y 2013-10-21 2013-12-20\n",
       "1           25             Y 2013-10-21 2013-11-22\n",
       "2           20             Y 2013-07-09 2013-11-16\n",
       "3           23             Y 2013-08-10 2013-11-15\n",
       "4           21             Y 2013-09-16 2013-10-18"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "campaign_data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## *Train Data:*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 78369 entries, 0 to 78368\n",
      "Data columns (total 5 columns):\n",
      "id                   78369 non-null int64\n",
      "campaign_id          78369 non-null int64\n",
      "coupon_id            78369 non-null int64\n",
      "customer_id          78369 non-null int64\n",
      "redemption_status    78369 non-null int64\n",
      "dtypes: int64(5)\n",
      "memory usage: 3.0 MB\n"
     ]
    }
   ],
   "source": [
    "train.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>id</th>\n",
       "      <th>campaign_id</th>\n",
       "      <th>coupon_id</th>\n",
       "      <th>customer_id</th>\n",
       "      <th>redemption_status</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>13</td>\n",
       "      <td>27</td>\n",
       "      <td>1053</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>13</td>\n",
       "      <td>116</td>\n",
       "      <td>48</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>9</td>\n",
       "      <td>635</td>\n",
       "      <td>205</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>13</td>\n",
       "      <td>644</td>\n",
       "      <td>1050</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>9</td>\n",
       "      <td>8</td>\n",
       "      <td>1017</td>\n",
       "      <td>1489</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   id  campaign_id  coupon_id  customer_id  redemption_status\n",
       "0   1           13         27         1053                  0\n",
       "1   2           13        116           48                  0\n",
       "2   6            9        635          205                  0\n",
       "3   7           13        644         1050                  0\n",
       "4   9            8       1017         1489                  0"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "78369"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>customer_id</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>customer_id</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>94</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             customer_id\n",
       "customer_id             \n",
       "1                     94\n",
       "3                     42\n",
       "4                     23\n",
       "5                     52\n",
       "6                     32"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# More than one data points of each customer(customer_id) :\n",
    "\n",
    "train.groupby('customer_id').agg({'customer_id':'count'}).head() # COUNT "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">   In train data, more than 1 data point is present for the same customer_id\n",
    "\n",
    ">   More than one type of coupon is applied by each customer over the campaigns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## *Customer Transaction Data:*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>date</th>\n",
       "      <th>customer_id</th>\n",
       "      <th>item_id</th>\n",
       "      <th>quantity</th>\n",
       "      <th>selling_price</th>\n",
       "      <th>other_discount</th>\n",
       "      <th>coupon_discount</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>2012-01-02</td>\n",
       "      <td>1501</td>\n",
       "      <td>26830</td>\n",
       "      <td>1</td>\n",
       "      <td>35.26</td>\n",
       "      <td>-10.69</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2012-01-02</td>\n",
       "      <td>1501</td>\n",
       "      <td>54253</td>\n",
       "      <td>1</td>\n",
       "      <td>53.43</td>\n",
       "      <td>-13.89</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>2012-01-02</td>\n",
       "      <td>1501</td>\n",
       "      <td>31962</td>\n",
       "      <td>1</td>\n",
       "      <td>106.50</td>\n",
       "      <td>-14.25</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>2012-01-02</td>\n",
       "      <td>1501</td>\n",
       "      <td>33647</td>\n",
       "      <td>1</td>\n",
       "      <td>67.32</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>2012-01-02</td>\n",
       "      <td>1501</td>\n",
       "      <td>48199</td>\n",
       "      <td>1</td>\n",
       "      <td>71.24</td>\n",
       "      <td>-28.14</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         date  customer_id  item_id  quantity  selling_price  other_discount  \\\n",
       "0  2012-01-02         1501    26830         1          35.26          -10.69   \n",
       "1  2012-01-02         1501    54253         1          53.43          -13.89   \n",
       "2  2012-01-02         1501    31962         1         106.50          -14.25   \n",
       "3  2012-01-02         1501    33647         1          67.32            0.00   \n",
       "4  2012-01-02         1501    48199         1          71.24          -28.14   \n",
       "\n",
       "   coupon_discount  \n",
       "0              0.0  \n",
       "1              0.0  \n",
       "2              0.0  \n",
       "3              0.0  \n",
       "4              0.0  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer_transaction_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>customer_id</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>customer_id</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>1048</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>419</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>705</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>220</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>792</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             customer_id\n",
       "customer_id             \n",
       "1                   1048\n",
       "2                    419\n",
       "3                    705\n",
       "4                    220\n",
       "5                    792"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# More than one data points of each customer(customer_id) :\n",
    "\n",
    "customer_transaction_data.groupby('customer_id').agg({'customer_id':'count'}).head() # COUNT "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">   In customer_transaction_data, more than 1 data point is present for the same customer_id.\n",
    "\n",
    ">   More than one transaction has been performed by each customer over the campaigns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Aggregrating transaction featurs for each customer_id(customer) :\n",
    "\n",
    "# More than one data points of each customer(customer_id) :\n",
    "\n",
    "customer_transaction_data_groupBy = customer_transaction_data.groupby('customer_id').agg({'selling_price':['mean', 'max', 'min', 'sum'],\n",
    "                                                     'other_discount':['mean', 'max', 'min', 'sum'],\n",
    "                                                      'coupon_discount':['mean', 'max', 'min', 'sum']})  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "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 tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>customer_id</th>\n",
       "      <th colspan=\"4\" halign=\"left\">selling_price</th>\n",
       "      <th colspan=\"4\" halign=\"left\">other_discount</th>\n",
       "      <th colspan=\"4\" halign=\"left\">coupon_discount</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>mean</th>\n",
       "      <th>max</th>\n",
       "      <th>min</th>\n",
       "      <th>sum</th>\n",
       "      <th>mean</th>\n",
       "      <th>max</th>\n",
       "      <th>min</th>\n",
       "      <th>sum</th>\n",
       "      <th>mean</th>\n",
       "      <th>max</th>\n",
       "      <th>min</th>\n",
       "      <th>sum</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>94.001842</td>\n",
       "      <td>445.25</td>\n",
       "      <td>12.11</td>\n",
       "      <td>98513.93</td>\n",
       "      <td>-16.250382</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-372.23</td>\n",
       "      <td>-17030.40</td>\n",
       "      <td>-2.019876</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-106.86</td>\n",
       "      <td>-2116.83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>102.864033</td>\n",
       "      <td>640.80</td>\n",
       "      <td>14.25</td>\n",
       "      <td>43100.03</td>\n",
       "      <td>-16.830430</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-312.03</td>\n",
       "      <td>-7051.95</td>\n",
       "      <td>-0.595084</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-89.05</td>\n",
       "      <td>-249.34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>103.617404</td>\n",
       "      <td>862.00</td>\n",
       "      <td>8.90</td>\n",
       "      <td>73050.27</td>\n",
       "      <td>-22.714227</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-463.06</td>\n",
       "      <td>-16013.53</td>\n",
       "      <td>-3.091546</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-142.48</td>\n",
       "      <td>-2179.54</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>154.423727</td>\n",
       "      <td>1330.05</td>\n",
       "      <td>14.25</td>\n",
       "      <td>33973.22</td>\n",
       "      <td>-13.305409</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-390.04</td>\n",
       "      <td>-2927.19</td>\n",
       "      <td>-0.404773</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-89.05</td>\n",
       "      <td>-89.05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>130.827146</td>\n",
       "      <td>1485.35</td>\n",
       "      <td>12.11</td>\n",
       "      <td>103615.10</td>\n",
       "      <td>-13.657917</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-216.21</td>\n",
       "      <td>-10817.07</td>\n",
       "      <td>-0.114684</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-71.24</td>\n",
       "      <td>-90.83</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  customer_id selling_price                            other_discount       \\\n",
       "                       mean      max    min        sum           mean  max   \n",
       "0           1     94.001842   445.25  12.11   98513.93     -16.250382  0.0   \n",
       "1           2    102.864033   640.80  14.25   43100.03     -16.830430  0.0   \n",
       "2           3    103.617404   862.00   8.90   73050.27     -22.714227  0.0   \n",
       "3           4    154.423727  1330.05  14.25   33973.22     -13.305409  0.0   \n",
       "4           5    130.827146  1485.35  12.11  103615.10     -13.657917  0.0   \n",
       "\n",
       "                    coupon_discount                        \n",
       "      min       sum            mean  max     min      sum  \n",
       "0 -372.23 -17030.40       -2.019876  0.0 -106.86 -2116.83  \n",
       "1 -312.03  -7051.95       -0.595084  0.0  -89.05  -249.34  \n",
       "2 -463.06 -16013.53       -3.091546  0.0 -142.48 -2179.54  \n",
       "3 -390.04  -2927.19       -0.404773  0.0  -89.05   -89.05  \n",
       "4 -216.21 -10817.07       -0.114684  0.0  -71.24   -90.83  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer_transaction_data_groupBy.reset_index(inplace=True)\n",
    "customer_transaction_data_groupBy.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## *Customer Demographs:*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "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>customer_id</th>\n",
       "      <th>age_range</th>\n",
       "      <th>marital_status</th>\n",
       "      <th>rented</th>\n",
       "      <th>family_size</th>\n",
       "      <th>no_of_children</th>\n",
       "      <th>income_bracket</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>70+</td>\n",
       "      <td>Married</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>46-55</td>\n",
       "      <td>Married</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>7</td>\n",
       "      <td>26-35</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>8</td>\n",
       "      <td>26-35</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>10</td>\n",
       "      <td>46-55</td>\n",
       "      <td>Single</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   customer_id age_range marital_status  rented family_size no_of_children  \\\n",
       "0            1       70+        Married       0           2            NaN   \n",
       "1            6     46-55        Married       0           2            NaN   \n",
       "2            7     26-35            NaN       0           3              1   \n",
       "3            8     26-35            NaN       0           4              2   \n",
       "4           10     46-55         Single       0           1            NaN   \n",
       "\n",
       "   income_bracket  \n",
       "0               4  \n",
       "1               5  \n",
       "2               3  \n",
       "3               6  \n",
       "4               5  "
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer_demographs.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 760 entries, 0 to 759\n",
      "Data columns (total 7 columns):\n",
      "customer_id       760 non-null int64\n",
      "age_range         760 non-null object\n",
      "marital_status    431 non-null object\n",
      "rented            760 non-null int64\n",
      "family_size       760 non-null object\n",
      "no_of_children    222 non-null object\n",
      "income_bracket    760 non-null int64\n",
      "dtypes: int64(3), object(4)\n",
      "memory usage: 41.7+ KB\n"
     ]
    }
   ],
   "source": [
    "customer_demographs.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0xed5455e5c8>"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAxEAAAD9CAYAAADDCfoCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3deZwkdX3/8dd7FxC5JOFQZEkW5BJRYVlQgwcCSwARNLoBVOQwrhpRQVFDFESIv+AREZUgKwISCQgIugEECYqAyrEcy01EQF1AEUUOUWB33r8/6jtsM/TM1Mz2bHX3vJ+PRz+mu+pb1d+u7qn6fup7yTYRERERERF1TWk6AxERERER0VsSRERERERExJgkiIiIiIiIiDFJEBEREREREWOSICIiIiIiIsYkQURERERERIzJhAURknaWdIekOyX9y0S9T0REREREDE/SSZIekHTzMOsl6cul3H6jpBmj7XNCgghJU4HjgF2AzYC9JW02Ee8VEREREREjOgXYeYT1uwAblccc4PjRdjhRNRHbAHfavsv2k8AZwB4T9F4RERERETEM25cBfxghyR7Aqa5cCawuaZ2R9rlcJzPYYl3g1y2vFwKvaE0gaQ5VpMOmzNhqmjaYoKxEdK+L7lvQdBYmtb9/4cubzkJERCxjFw+cpabzUMfAbzZ23bRT1/n5eyjl6mKu7bljeLt2Zfd1gfuH22Cigoh2X84zDkT5YHMBZk2ZXfsgRfSTFGIjogm5gdGsnPuj01rL1eM0atl9qIkKIhYC67W8ngbcN0HvFRERERHRVwYYqJ22A/0Txlx2n6g+EdcAG0laX9IKwF7AvAl6r4iIiIiIvvKUF9d+dMA84J1llKZXAg/bHrYpE0xQTYTtRZIOBC4CpgIn2b5lIt4ropelSUGz0qQgIiK61VhqIkYj6XRgO2BNSQuBTwHLA9j+GnABsCtwJ/A4sP9o+5yo5kzYvqBkKCKGkUJsREREtLPYnesybHvvUdYbeP9Y9jlhQURERER0p9zAiOh+AyP3a25cgoiIiIhJJk0pm5UgLupY3K9BhKQVgcuA55T9nG37U5IuB1YtydYGrrb9pqXOaUQfyoW8WbmQR0REt+rnmogngO1tPyZpeeAKSd+3/ZrBBJK+A3xvaTMZ0a9SiI2IiIh2nupgn4iJMO4gonTAeKy8XL48nv60klYFtqdG7+6IiIhYdnIDI6L79W1zJgBJU4FrgQ2B42xf1bL6zcAlth8ZZts5lOm5N2UG07TB0mQloielOVOzUpCKySrnnmbl3BN1LO7uGGLpggjbi4EtJK0OnCtpc9s3l9V7AyeOsO3T03PPmjK7yw9TxMTIhSQiIiLa6dwsEROjIzNW2/4jcCmwM4CkNYBtgPM7sf+IiIiIiMlkMar9aMK4gwhJa5UaCCQ9F9gRuL2sng2cZ/svS5/FiIiIiIjJZcD1H01YmuZM6wDfLP0ipgBn2j6vrNsLOHppMxfR79IuuVlpThYREd3qyc40GJowSzM6043AlsOs2268+42IiIiImOwG3EwzpboyY3VEg3InPCKakHNPRPdrqq9DXQkiIiIiJpk0pWxWgrioY3GXN2daqtxJ2kTSDS2PRyQd1LL+EEmWtObSZzUiIiI6IYXYiO43YNV+NGFp54m4A9gCnp547l7g3PJ6PWAW8KulzGNE38rdwGalIBWTVc49Ed3vSU9tOgsj6mRzph2AX9j+ZXl9DPAx4HsdfI+IvpJCbERERLQz0OXNmToZROwFnA4gaXfgXtsLpPZVLJLmAHMANmUG07RBB7MS0RtyN7BZCeIiIqJbdXvHatlLP0OFpBWA+4CXAI8CPwJ2sv2wpHuAmbYfHG77WVNmNzRNRkRERERMJhcPnNXdpfPi4rtfXLt8PGv925b5Z+pUTcQuwHW2fyvppcD6wGAtxDTgOknb2P5Nh94voi+kJqJZqYmIySrnnmbl3BN1DHR5TUSngoi9KU2ZbN8ErD24ok5NRERERERELNHtQ7wudRAhaSWqUZjes/TZiZhccjcqIiIi2nnK3T2d21LnzvbjwBojrJ++tO8R0a/SpKBZCeIiIqJbLW5o/oe6ujvEiehzKcRGRBNy7onofn3fnCkixi81Ec1KQSoiIrrVgHs8iJB0ErAb8IDtzcuy2cARwIuBbWzPL8vXAM4GtgZOsX3gBOU7oi+kEBsRTcgNjGbl3B919ENNxCnAV4FTW5bdDPwDcMKQtH8BDgM2L4+IiIjoMinERnS/nu8TYfsySdOHLLsNYOhs1Lb/BFwhacPOZTGif+VuYLNSkIrJKueeZuXcE3V0++hMjdWTSJojab6k+Qt9V1PZiIiIiIjoOgOo9qMJjQURtufanml75jRt0FQ2IiIiIiK6zmJPqf1oQnfXk0T0uVRpR0RERDv90LE6IiIiIiKWoYFe71gt6XRgO2BNSQuBTwF/AL4CrAWcL+kG239f0t8DrAasIOlNwE62b52Y7EdERERE9J9u71hdZ3SmvYdZde4w6acvTYYiIiIiIia7xQ11mK6ru0OciIiIiIhJqOdnrI6IiZOx2puVju0REdGt+qImQtJJwG7AA7Y3H7LuEODzwFq2H5T0POBbwN+U/X/B9smdzXZEf0ghNiKakHNPRPfrZE2EpJ2BY4GpwIm2jx6y/m+AbwKrlzT/YvuCkfZZtybiFOCrwKlD3nA9YBbwq5bF7wdutf1GSWsBd0g6zfaTNd8rYtJITUSzUpCKiIhu1an5HyRNBY6jKrMvBK6RNG/IwEefBM60fbykzYALgOkj7bdWEGH7MkntdnQM8DHge63JgVUlCViFaiSnRXXeJyIiIiZebmA0Kzcwoo6nPLVTu9oGuNP2XQCSzgD2AFqDCFONrgrwPOC+0XY67hBH0u7AvbaHnom+Cry4vPlNwIdsD7TZfo6k+ZLmL6w+U0REREREUM0TUffRWq4ujzktu1oX+HXL64VlWasjgHeU6RwuAD4wWv7G1bFa0krAJ4Cd2qz+e+AGYHvgRcDFki63/UhrIttzgbkAs6bM9njyERERERHRj8YyY3VrubqNdj20h5a99wZOsf0fkl4F/JekzdtVBAwab03Ei4D1gQVlcrlpwHWSXgDsD5zjyp3A3cCm43yfiIiIiIhJZyw1EaNYCKzX8noaz26u9C7gTADbPwNWBNYcaafjCiJs32R7bdvTy+RyC4EZtn9D1cl6BwBJzwc2AdJeKSIiIiKipgGm1H6M4hpgI0nrS1oB2AuYNyRNa/n9xVRBxO9G2mmtIELS6cDPgE0kLZT0rhGSHwX8naSbgEuAj9t+sM77REREREQELLZqP0ZiexFwIHARcBvVKEy3SDqy9HEG+AjwbkkLgNOB/WyP2N2g7uhMe4+yfnrL8/to31ciIiIiIiJqWDTQsdGZKHM+XDBk2eEtz28Fth3LPjNjdUREREREl+mLGasjYmJkrPCIiIhop0aH6UaNGkRIOgnYDXjA9uZl2RbA16g6XSwC/tn21ZK2o5p47u6y+Tm2j5yIjEdERMT45AZGRPcb6NCM1ROlTk3EKVQTyJ3asuxzwKdtf1/SruX1dmXd5bZ362QmIyIionMyY3WzEsRFHQO93pzJ9mWSpg9dzBinxo6IiIjukEJsRPd7qoMdqyfCePtEHARcJOkLVMPE/l3LuleV4aHuAw6xfUu7HZTpuOcAbMoMpmmDcWYlIiIiIqK/dHufiPE2tnofcLDt9YCDgW+U5dcBf2v75cBXgO8OtwPbc23PtD0zAURERERExBIDqPajCeMNIvYFzinPzwK2AbD9iO3HyvMLgOUljThldkREREREPNOAVfvRhPE2Z7oPeB1wKbA98HMASS8AfmvbkrahClJ+34F8RvSldG5sVtqFx2SVc0+zcu6JOnp+dCZJp1ONvLSmpIXAp4B3A8dKWg74C6VvA/BW4H2SFgF/BvYabcrsiMksF5KIaELOPRHdr9v7RNQZnWnvYVZt1SbtV6mGg42IiIiIiHFa1Os1ERERERERsWz1fE2EpPWoJpp7ATAAzLV9rKTPA28EngR+Aexv+4+SZgFHAyuUdR+1/cOJ+gARvSztkpuVJh0xWeXc06yce6KOng8igEXAR2xfJ2lV4FpJFwMXA4faXiTps8ChwMeBB4E32r5P0ubARcC6E5T/iJ6WC0lENCHnnoju1/NBhO37gfvL80cl3Qasa/sHLcmupOpUje3rW5bfAqwo6Tm2n+hctiMiIiIi+ldT8z/UNaY+EZKmA1sCVw1ZdQDw7TabvAW4PgFERERE90hzpmalJijq6PmaiEGSVgG+Axxk+5GW5Z+gavJ02pD0LwE+C+w0zP7mUIaG3ZQZZNbqiIiIiIjKooE+GJ1J0vJUAcRpts9pWb4vsBuwQ+t8EJKmAecC77T9i3b7tD0XmAswa8rszCURERGxjOROeET36/maCEkCvgHcZvuLLct3pupI/Trbj7csXx04n6rT9U86n+WI/pEmBc1KQSomq5x7mpVzT9ThXg8igG2BfYCbJN1Qlv0r8GXgOcDFVZzBlbbfCxwIbAgcJumwkn4n2w90NOcRERExLinERnS/nu9YbfsKaPspLhgm/b8B/7aU+YqYFHIhj4gmpCaiWTn3Rx0935wpIiIiIiKWrcX90LE6IiZG7gY2K3cDIyKiW/V8nwhJ6wGnAi8ABoC5to+VdATwbuB3Jem/2r5A0grACcDMkv5Dti+dgLxH9LwUYiMiIqKdfmjOtAj4iO3rJK0KXCvp4rLuGNtfGJL+3QC2XyppbeD7kra2PdC5bEf0h9RENCtBXEREdCt3+QQIoza2sn2/7evK80eB24B1R9hkM+CSkv4B4I9UtRIREREREVHDAKr9aMKY+kRImg5sCVxFNfTrgZLeCcynqq14CFgA7CHpDGA9YKvy9+rOZTsiIiLGK7VwEd2v5/tEDJK0CtWs1QfZfkTS8cBRgMvf/wAOAE4CXkwVWPwS+ClVk6ih+5sDzAHYlBlM0wZL90kielAu5BHRhDSlbFbO/VHH4oE+CCIkLU8VQJxm+xwA279tWf914LyyfBFwcMu6nwI/H7pP23OBuQCzpszu8lZfERERERHLTs/XRKiajvobwG22v9iyfB3b95eXbwZuLstXAmT7T5JmAYts39r5rEdERERE9KeeDyKo+j7sA9wk6Yay7F+BvSVtQdWc6R7gPWXd2sBFkgaAe8u2EdFGmhQ0K00KIiKiW/X8EK+2r4C23b4vGCb9PcAmS5etiMkhhdiIaELOPRHdr9uHeM2M1RENSk1Es1KQiskq555m5dwTdQwMjDoTQ6MSREREREwyKcRGdL8ur4io1bF6PeBU4AXAADDX9rGSXg58DViFqk/E220/UrZ5GXACsFrZZmvbf5mQTxDRw3Ihj4gmpCaiWTn3Rx2d7FgtaWfgWGAqcKLto9uk+UfgCKr4ZYHtt420zzo1EYuoJpK7TtKqwLWSLgZOBA6x/WNJBwAfBQ6TtBzwLWAf2wskrQE8VftTRkRExIRKITaiB3SoKkLSVOA4YBawELhG0rzW0VMlbQQcCmxr+yFJa4+23zodq+8H7i/PH5V0G7AuVefpy0qyi4GLgMOAnYAbbS8o2/y+9qeMmGRyN7BZKUjFZJVzT7Ny7ok6OlgTsQ1wp+27ACSdAewBtE7B8G7gONsPVe/tB0bb6Zh6bEiaDmwJXEU1L8TuZdVsYL3yfGPAki6SdJ2kjw2zrzmS5kuav7D6TBERERERQTU6U91Ha7m6POa07Gpd4NctrxeWZa02BjaW9BNJV5bmTyOq3bFa0ipUs1YfZPuR0oTpy5IOB+YBT7bs89XA1sDjwCWSrrV9yTMPTGasjsjdqIiIiGjHYxidqbVc3Ua7Ko2hZe/lgI2A7YBpwOWSNrf9x+Hes1YQIWl5qgDiNNvnlMzeTtV0CUkbA28oyRcCP7b9YFl3ATADuGTofiMmuzQpaFaCuIiI6FYdnCdiIUtaDEEVJNzXJs2Vtp8C7pZ0B1VQcc1wO60zOpOAbwC32f5iy/K1bT8gaQrwSaqRmqDqG/ExSStR1U68DjhmtPeJmIxSiI2IJuTcE9EDOhdEXANsJGl94F5gL2DoyEvfBfYGTpG0JlXzphH7G9SpidgW2Ae4SdINZdm/lsy8v7w+BzgZoPTo/mLJsIELbJ9f430iIiIiIoLOday2vUjSgVQ3+qcCJ9m+RdKRwHzb88q6nSTdCiwGPjra4EhyF8ypnT4RMVmlOVOzcjc2Jquce5qVc0+zLh44q3MTMEyg9b/177XLx3e/49Bl/pkyY3VERERERJfp5GRzEyFBRESDcjcqIpqQc09ED+jyIGLUsaMkrSjpakkLJN0i6dNl+TfKshslnV2GgEXSeyXdJOkGSVdI2myiP0RERERERF/xGB4NqFMT8QSwve3HylCvV0j6PnCw7UcASkfqA4Gjgf+2/bWyfHfgi8CoE1ZERETEspE+Ec1KTVDU0uU9hkcNIlz1vH6svFy+PNwSQAh4LuWjDi4vVqbrD0FEc3Ihb1Yu5BER0bW6vDlT3cnmpgLXAhsCx9m+qiw/GdgVuBX4SEv69wMfBlYAth9mn3OAOQCbMoNp2mD8nyKiR6UQGxEREe10wQCqI6oVRNheDGwhaXXg3DIN9s229y8BxleAPVkyV8RxwHGS3kY1Ed2+bfb59PTcGeI1JqvURDQrQVxERHStgT6oiRhk+4+SLqXq43BzWbZY0reBj1KCiBZnAMd3IJ8RfSmF2IhoQs49Ed1PXX6LfdQgQtJawFMlgHgusCPwOUkb2r6z9Il4I3B7Sb+R7Z+Xzd8A/LztjiMiNRENS0EqIiK6Vq8HEcA6wDdLs6UpwJnA+cDlklYDBCwA3lfSHyhpR+Ap4CHaNGWKiEoKsREREdFWr3estn0jsGWbVdsOk/5DS5upiIiImDipBW1WbiBFLX1QExEREREREctSgoiIiIjoJrkTHtEDen10JkkrApcBzynpz7b9KUmXA6uWZGsDV9t+U+lofSzV/BGPA/vZvm5Cch8RERFjluZMzUoQF3X0/OhMwBPA9rYfk7Q8cIWk79t+zWACSd8Bvlde7gJsVB6voBri9RWdzXZEf8iFvFm5kEdERNfq9SDCtoHHysvly+PpjyVpVapZqfcvi/YATi3bXSlpdUnr2L6/ozmP6AMpxEZEREQvqtUnogzvei2wIXCc7ataVr8ZuMT2I+X1usCvW9YvLMueEURImgPMAdiUGUzTBuP6ABG9LDURzUoQFxER3arbmzNNqZPI9mLbWwDTgG0kbd6yem/g9JbX7XqBPOsw2J5re6btmQkgIiIiIiJaWPUfDRjT6Exl1upLgZ2BmyWtAWxDVRsxaCGwXsvracB9S5nPiL6UO+ER0YSceyJ6wEDTGRhZndGZ1gKeKgHEc4Edgc+W1bOB82z/pWWTeVSzVp9B1aH64fSHiGgvzZmalYJUTFY59zQr556oo9ubM9WpiVgH+GbpFzEFONP2eWXdXsDRQ9JfQDW8651UQ7zuT0RERHSNFGIjekCvBxG2bwS2HGbddm2WGXj/UucsIiIiImKy6vUgIiIiIvpLmjM1KzVBUUc/NGeKiIiIiIhlaaCZUZfqqh1ElD4R84F7be8maX3gDOCvgeuAfWw/KekY4PVls5WAtW2v3uF8R/SF3I2KiIiIdvqpJuJDwG3AauX1Z4FjbJ8h6WvAu4DjbR88uIGkDzBMf4qISJOCpiWIi8kqv/2IHtAPQYSkacAbgM8AH5YkYHvgbSXJN4EjgOOHbLo38KmO5DQiIiI6IjcwmpUgLuro9pqIWjNWA18CPsaSaS/WAP5oe1F5vRBYt3UDSX8LrA/8sN0OJc2RNF/S/IW+a8wZj4iIiIjoWx7DowF1JpvbDXjA9rWSthtc3Cbp0I+wF3C27cXt9mt7LjAXYNaU2V0ea0VMjNyNioiIiLa6vHRcpznTtsDuknYFVqTqE/ElYHVJy5XaiGnAfUO224vMFxERERERMWbd3pypzmRzhwKHApSaiENsv13SWcBbqUZo2hf43uA2kjYB/gr42QTkOaJvpF1ys1ITFBERXavXg4gRfBw4Q9K/AdcD32hZtzdwRpm9OiKGkUJsRDQh556I7tfzNRGtbF8KXFqe3wVsM0y6I5YyXxERETFBUgvarARxUUs/BRER0Vm5kDcrF/KIiOhaCSIiIiIiImIs+qY5k6SpwHzgXtu7tSz/CrC/7VWGpH8rcBawte35HcpvRF/JnfCIaELOPRHdTwOjp6m9L2ln4FhgKnCi7aOHSVe7/F53sjmADwG3DXmjmcDqbTKwKvBB4Kox7D8iIiIiIqBjk82VioDjgF2AzYC9JW3WJt2Yyu+1aiIkTQPeAHwG+HBLhj4PvA1485BNjgI+BxxSZ/8RERGx7KQ/VrNSExS1dK450zbAnWVQJCSdAewB3Dok3ZjK73VrIr4EfAxorVg5EJhn+/7WhJK2BNazfd5IO5Q0R9J8SfMXVp8pIiIiIiKo+kTUfrSUq8tjTsuu1gV+3fJ6YVm25L1qlt9bjVoTIWk34AHb15bJ5pD0QmA2sN2QtFOAY4D9Rtuv7bnAXIBZU2Z3edeRiIiI/pE74RE9YAyl49ZydRsaae9jKb+3qtOcaVtgd0m7AisCqwG3AE8Ad0oCWEnSncBWwObApWX5C4B5knZP5+qIZ0uTgmalIBWTVc49zcq5J+roYMfqhcB6La+nAfe1vF6VcZTfR23OZPtQ29NsTwf2An5o+69sv8D29LL8cdsb2n7Y9poty68EEkBERERERIxFhzpWA9cAG0laX9IKVOX5eU+/zTjL75knIqJBuRsVERER7XRqngjbiyQdCFxENcTrSbZvkXQkMN/2vJH30N6YggjblwKXtlm+yrMSV8u3G0+mIiIiIiImtQ72GLZ9AXDBkGWHD5N2uzr7TE1ERETEJJNa0Ige0OXDDiWIiIiImGTSsbpZCeKijnZDKnWT2kFEmVxuPnCv7d0k7UA12dwU4DFgP9t3Snot1bwSLwP2sn32BOQ7IiIiximF2Iju18HRmSbEWGoiPgTcRjXEK8DxwB62b5P0z8AnqcaX/VX5m9mqIyIiulBqIpqVIC5q6YfmTJKmAW8APgN8uCw2SwKK51HGm7V9T9mmy+OniIiIySmF2Ige0A9BBFXzpI9RTUYx6J+ACyT9GXgEeOVY3rhMxz0HYFNmME0bjGXziIiIGKfURDQrQVzU0akhXifKqJPNSdoNeMD2tUNWHQzsansacDLwxbG8se25tmfanpkAIiIiYtlJITaiB3RusrkJUacmYltgd0m7AisCq0k6H9jU9lUlzbeBCycojxF9K3cDm5WCVExWOfdEdL9u7xgwahBh+1DgUABJ21F1mH4T8BtJG9v+P2AWVafriBiDFGIjIiKinW5vzjSueSLK9NnvBr5TOlA/BBwAIGlr4Fzgr4A3Svq07Zd0KsMRERGxdHIDI6IH9FMQYftS4NLy/FyqYGFommuAaR3IW0TfS5OCZqUgFZNVzj3NyrknaumnICIiIiIiIiZeXzRnknQP8CiwGFhke6ak2cARwIuBbWzPL2mXB04EZpT9n2r73zuf9Yjel7tREdGEnHsiekA/BBHF620/2PL6ZuAfgBOGpJsNPMf2SyWtBNwq6fTBSegiIiIiImJkGujuKGLczZls3wYg6VmrgJUlLQc8F3iSajK6iIiI6ALpE9Gs1ARFHX3RnIkqMPiBJAMn2J47QtqzgT2A+4GVgINt/2HpshnRn3Ihb1Yu5BER0bX6JIjY1vZ9ktYGLpZ0u+3Lhkm7DVXfiRdSDfN6uaT/tX1XayJJc4A5AJsyg8xaHZNRCrER0YSceyK6X7fXREypk8j2feXvA1TDum4zQvK3ARfafqqk/wkws80+59qeaXtmAoiIiIiIiBYew6MBo9ZESFoZmGL70fJ8J+DIETb5FbC9pG9RNWd6JfClTmQ2IiIill6aUjYrNUFRRz/URDwfuELSAuBq4HzbF0p6s6SFwKuA8yVdVNIfB6xCNXrTNcDJtm+cgLxHRERERPQlDdR/NGHUmojSl+FZIfMIM1Y/RjXMa0SMIncDm5W7gRER0bXc3VURmbE6okEpxEZEREQ73d6cKUFERINSE9GsBHEREdG1ujyIqDU6k6R7JN0k6QZJ88uyIyTdW5bdIGnXsny6pD+3LP/aRH6AiIiIiIh+0/N9Ilq83vaDQ5YdY/sLbdL+wvYWS5GviIiImCCphYvofk0FB3WlOVNERERERLfpk47VBn4gycAJtueW5QdKeicwH/iI7YfK8vUlXQ88AnzS9uUdzXVEn8jdwIhoQvpjNSvn/qijXzpWb2v7PklrAxdLuh04HjiKKsA4CvgP4ADgfuBvbP9e0lbAdyW9xPYjrTuUNAeYA7ApM8is1TEZ5ULerFzIIyKia/VDEGH7vvL3AUnnAtvYvmxwvaSvA+eVNE8AT5Tn10r6BbAxVW1F6z7nAnMBZk2Z3eWHKWJipBAbERER7fR8TYSklYEpth8tz3cCjpS0ju37S7I3U81QjaS1gD/YXixpA2Aj4K6JyX5ERESMVW5gRPSAPugT8XzgXEmD6f/b9oWS/kvSFlSVLfcA7ynpX0sVZCwCFgPvtf2Hjuc8IiIixiVNKZuVIC7q6PnRmWzfBTzr1257n2HSfwf4ztJnLSIiIiJicur55kwRMXFyN7BZuRsYERFda6C7o4gEERERERER3aa7Y4h6QYSke4BHqfo4LLI9U9LLga8Bq1D1iXi77UckzQKOBlYAngQ+avuHE5D3iIiIGIfUwkV0v042Z5K0M3AsMBU40fbRQ9Z/GPgnYBHwO+AA278caZ9jqYl4ve0HW16fCBxi+8eSDgA+ChwGPAi8scwrsTlwEbDuGN4nIiIiJlCaUjYrQVzU0qHRmSRNBY4DZgELgWskzbN9a0uy64GZth+X9D7gc8CeI+13ylLkaRNgcK6Ii4G3ANi+fnBeCeAWYEVJz1mK94mIiIiImFQ0UP8xim2AO23fZftJ4Axgj9YEtn9k+/Hy8kpg2mg7rVsTYeAHkgycUCaKuxnYHfgeMBtYr812bwGuLxPQPUNmrI7I3aiIaEbOPRHdT56q2mUAABgeSURBVGOoiWgtVxdzS3kdqhZBv25ZtxB4xQi7exfw/dHes24QsW1pnrQ2cLGk24EDgC9LOhyYR9X/4WmSXgJ8lmpyumfJjNURaVLQtBSkYrLKuadZOfdELWOYJ6K1XN2G2m3SNqH0DmAm8LrR3rNWEDHYPMn2A5LOBbax/QVKgCBpY+ANLRmYBpwLvNP2L+q8R8RklAtJRDQh556I7jeWmohRLOSZLYamAfcNTSRpR+ATwOvatSIaatQgQtLKwBTbj5bnO1HNSL12CSqmAJ+kGqkJSasD5wOH2v7J6J8rYvLK3cBmpSAVk1XOPc3KuSdq6Vw7nWuAjSStD9wL7AW8rTWBpC2BE4CdbT9QZ6d1aiKeD5wraTD9f9u+UNKHJL2/pDkHOLk8PxDYEDhM0mFl2U51MxQxmeRCEhEREe2oQ5PN2V4k6UCqEVOnAifZvkXSkcB82/OAz1NN23BWKfP/yvbuI+131CDC9l3As0o6to+lGm926PJ/A/5t9I8UEbkb2KwEcRER0bU615wJ2xcAFwxZdnjL8x3Hus/MWB3RoBRiIyIiop0aQ7c2KkFERERERES36WBNxESoNdmcpNUlnS3pdkm3SXqVpM+X1zdKOrd0qEbSCpJOlnSTpAWStpvQTxARERER0W88hkcD6tZEHAtcaPutklYAVqKapfrQ0lnjs8ChwMeBdwPYfmmZV+L7kra23eWVMhHLXvpENCvNySIiolt1cIjXCVFniNfVgNcC+wGU6bKfBH7QkuxK4K3l+WbAJSXtA5L+SDVpxdUdy3VEn0ghNiIiItpa3N1BRJ3mTBsAvwNOlnS9pBPLfBGtDmDJ9NgLgD0kLVfGo92KZ05wAVTTc0uaL2n+Qt+1FB8hIiIiIqK/yK79aEKd5kzLATOAD9i+StKxwL8AhwFI+gSwCDitpD8JeDEwH/gl8NOy/hlap+eeNWV2d4daERERfSS1oBE9oNebM1FNlb3Q9lXl9dlUQQSS9gV2A3awq09qexFw8ODGkn4K/LyTmY6IiIiI6GtdHkSM2pzJ9m+AX0vapCzaAbhV0s5UHal3t/34YHpJKw02d5I0C1hk+9bOZz0iIiIiok8NjOHRgLqjM30AOK2MzHQXsD9wDfAc4OIyPfaVtt8LrA1cJGkAuBfYp+O5jugTGZ2pWWnSEZNVzj3Nyrkn6uj50ZkAbN9ANcJSqw2HSXsPsEm7dRHxTLmQRERERFsD3T07QmasjoiImGRyAyOiB/RDTURERET0jzRnalaCuKiluysi6gURklYHTgQ2p5pc+wDbP5P0AeBAqiFcz7f9sZL+ZcAJwGpUh2Br23+ZgPxHRETEGKUQG9H9+qJPBHAscKHtt5bO1StJej2wB/Ay209IWhtA0nLAt4B9bC+QtAbw1ERkPiIiIsYuNRHNShAXtfR6ECFpNeC1wH4Atp8EnpT0PuBo20+U5Q+UTXYCbrS9oCz//QTkO6Iv5ELerFzIIyKiay3u7vZMdWoiNgB+B5ws6eXAtcCHgI2B10j6DPAX4BDb15TllnQRsBZwhu3PDd2ppDnAHIBNmcE0bdCJzxPRU1KIjYiIiLZ6vSaipJkBfMD2VZKOpZqxejngr4BXAlsDZ0raoCx/dVn2OHCJpGttX9K6U9tzgbkAs6bM7u6jFDFBUhPRrARxERHRtfogiFgILLR9VXl9NlUQsRA4x7aBq8vkcmuW5T+2/SCApAuogpBLnrXniEkuhdiIiIhoa6C7g4gpoyWw/Rvg15IGJ5DbAbgV+C6wPYCkjYEVgAeBi4CXSVqpdLJ+XUkfERERERF1eKD+owF1R2f6AHBaGZnpLmB/4E/ASZJuBp4E9i21Eg9J+iJwDdVwsBfYPr/zWY/ofWnO1KzUBMVkld9+RA/og+ZM2L4BmNlm1TuGSf8tqmFeI2IEuZBHREREW30wOlNERET0kdSCNis3kKKWfqiJiIiJkQt5s3Ihj8kqv/2IHtDrQUTpUP3tlkUbAIcDp5bl04F7gH+0/ZCkPYCjgAFgEXCQ7Ss6m+2I/pALeURERLQ10OPNmWzfAWwBIGkqcC9wLtUwr5fYPlrSv5TXH6caynWebUt6GXAmsOkE5T+i56U2ojkJ4mIyy7mnOTn3RC29XhMxxA7AL2z/stQ4bFeWfxO4FPi47cda0q9MNUJTRAwjF5OIaELOPRFdrsuDiFHniRhiL+D08vz5tu8HKH/XHkwk6c2SbgfOBw5otyNJcyTNlzR/oe8ae84jIiIiIvqUFy+u/WhC7ZqIMkfE7sCho6W1fS5wrqTXUvWP2LFNmrnAXIBZU2Z3d6gVMUHSnKBZuRMbk1XOPc3KuSdq6fUZq1vsAlxn+7fl9W8lrQNQ/j4wdAPblwEvkrTmUuc0IiIiImKysOs/GjCWIGJvljRlApgH7Fue7wt8D0DShpJUns8AVgB+v/RZjYiIiIiYJAYG6j8aUKs5k6SVgFnAe1oWHw2cKeldwK+A2WX5W4B3SnoK+DOwp93lPUMiGpIq7YiIiGiry4vPtYII248DawxZ9nuq0ZqGpv0s8NmO5C4iIiIiYhJqqsN0XZmxOqJB6dzYrNQERURE1+ryjtUJIiIalEJsREREtOUen7Fa0ibAt1sWbQAcbvtLZf0hwOeBtWw/WDpVHwvsCjwO7Gf7uo7nPCIiIsYlNzAiup87WBMhaWeq8vlU4ETbRw9Z/xzgVGArqgGR9rR9z0j7HDWIsH0HsEV5g6nAvcC55fV6VB2uf9WyyS7ARuXxCuD48jcihkhzpmalIBUREV2rQzURpfx+HFWZfSFwjaR5tm9tSfYu4CHbG0rai6p/854j7XeszZl2AH5h+5fl9THAxyjDuxZ7AKeWEZmulLS6pHUGZ7eOiCVSiI2IiIh2OlgTsQ1wp+27ACSdQVVebw0i9gCOKM/PBr4qSSONsDrWIGIvylwRknYH7rW9oEwLMWhd4NctrxeWZc8IIiTNAeaUl+8pM1j3JElzejn/vS7Hvzk59s3K8W9Wjn9zcuybleO/bFy8+NsaPVVlSLkaYG7Ld9SubD60ldDTaWwvkvQw1cisDw73nrUnm5O0ArA7cFaZN+ITwOHtkrZZ9qwoxvZc2zPLo9d/iHNGTxITKMe/OTn2zcrxb1aOf3Ny7JuV499lhpSrh5at65TNa5XfW41lxupdgOts/xZ4EbA+sEDSPcA04DpJL6CKbtZr2W4acN8Y3iciIiIiIjqjTtn86TSSlgOeB/xhpJ2OJYjYm9KUyfZNtte2Pd329PLGM2z/BphHNWO1JL0SeDj9ISIiIiIiGnENsJGk9UvLor2oyuut5gH7ludvBX44Un8IqNknojRfmgW8p0byC6iGd72TaojX/eu8R4/r9eZYvS7Hvzk59s3K8W9Wjn9zcuyblePfQ0ofhwOBi6iGeD3J9i2SjgTm254HfAP4L0l3UtVA7DXafjVKkBEREREREfEMY2nOFBERERERkSAiIiIiIiLGJkFE9LwyikDEpCRpO0l/N47t7pG05kTkKSIi+l+CiCEk/WvTeQCQ9EJJZw+z7lJJM5d1nsZD0nclXSvpljIRCpLeJen/yuf4uqSvluVrSfqOpGvKY9sR9nuEpLmSfgCcKmm6pMslXVcef1fSbVfe52xJt0s6TWV2REm7lmVXSPqypPPK8pUlnVTycL2kPSb8QDWo9bcmaQtJu9bYZrvB4zWG95ku6W2dStePyqh2Yz0vbweMOYjoV5I+KOk2Sact5X6OlLRjed6xc66kEyVt1ol99RNJm0q6oZxzXzSG7U6R9NY2y1vPa8Oer7o5mJb006bzMJrhjv+y3q+k/SS9sNP5iJEliHi2ZRpEDHcX3fZ9tjv+j9mAA2xvBcwEPihpXeAw4JVUI35t2pL2WOAY21sDbwFOHGXfWwF72H4b8AAwy/YMYE/gyy3ptgQOAjYDNgC2lbQicAKwi+1XA2u1pP8E1dBmWwOvBz4vaeWxf/TuJ2m5Ib+1LahGV5sI04E6wUHddH2hBE23SfpP4DpgH0k/K8HwWZJWKenukfTpsvymUuiaDrwXOLgUwF4zXDAuaQ1JPyiFtBNoP7FQP/hnYFfbb1+andg+3Pb/dihPrfv9J9u3dnq/feBNwPdsb2n7F0u7s6W5ho4zmO84231xc2AZtRbYD0gQsYw1/k/SaZLeKelGSQsk/dfQaFbSY+XvOpIuKxfem8vF92jguWXZaSXdh8v6myUdVJZNL3ewTyzLT5O0o6SfSPq5pG1KurZ3tEvEfJak/wF+MMznmC7p5vL8uZLOKJ/r28BzJ/AQdtoHJS0ArqSaxGQf4Me2/2D7KeCslrQ7Al+VdAPVeMWrSVp1hH3Ps/3n8nx54OuSbir7bL3Td7XthbYHgBuoCqmbAnfZvrukOb0l/U7Av5R8XAqsCPzNGD/3hKrzGyyPn5bf3k8lbVK2fcbvb/C3pmrs6COBPcv/wJ7D7aNG/l5X9jF4Z3FV4GjgNWXZwRqm9qhNuv1UaqvKvs9TdWdxavn/vrkUqg/u6EFetjYBTqUKrN8F7FgC4vnAh1vSPViWHw8cYvse4GtUwfcWti9n+GD8U8AVtrek+v/qqt90J0j6GtWNgnmSPj7C7/+7kv5H0t2SDizn+eslXSnpr0u6Z90JVVWLekzL63dL+uIweVlZ0vnlWnSzpD3L8kslzZS0e8v/yB2S7i7rt5L0Y1U1uBdJWmdijtbYaUnA+3VVtcs/KNenLcqxu1HSuZL+aoR9PCutqtrPg4B/kvSjEbZ9xvW9ZdVry3d81+B31noNHbKPtsG0nh3MrydpJ9UM6MdzPEejJeWVkWrUty6ffYGkqyWtKmlFSSeXvF0v6fUlbd3f/oskXVh+g5fX+Hw7lnT/J2m3lvdqvc6sIumSlmP2dA3/CN/r4Pqjyv/jlHb/H+U7nwmcVv6feqmM1Nts980DeAlwB7Bmef3XwCnAW1vSPFb+fgT4RHk+FVi1dX15vhVwE7AysApwC9Vd7enAIuClVIHYtcBJVCejPYDvlu3/H/CO8nx14P/KvvajmqDvr0f4LNOBm8vzD1ON6QvwsvLeM5s+3jW+j+2AK4CVyutLgTcD32xJ80Hgq+X5g8Bza+77CKpCVOvrL5TvYzlgUUsezmtJ99Vy/LekCmYGl+8+mK58n5s0ffxG+fyj/gaB1YDlSvodge+U58/4/Q35re03+H2U18Pt4xnHtU3+/gfYtjxfpXwnQ7+LlYAVy/ONqMaqbvedDc3TeSXNVsDFLctXb/p7WYrv8u7yfLfyf3BDedwKfKOsuwdYtzx/BfC/w/wvPNCy/Q3AvcCq5fkGLen+QDlX9tOjHKc1R/n931mOyVrAw8B7y7pjgIPK81Mo1w6qc9dMqvP3L4Dly/KfAi8dJh9vAb7e8vp5rfsakvZM4P1UN0N+CqxVlu9JOfd3w4Ml550tWvL9DuBG4HVl2ZHAl0bYR9u0Q3/HbbZ71vW95Xs6i+o8uBlwZ0teB89r27Hk/P5l4PDy/A2Ay+9lOjAAvLKsWxO4DFi5vP54y3b3AB8oz/8ZOHGCjvdjLfl/mGqW4SnAz4BXAysAdwFbl3SrUZ1rPwKcXJZtCvyK6mbYftT77V8CbFSev4KqZn64PJ4CXFjytRHVtWXwvVqvM8sBq7Uc2zuprlcjfa9vBT5H1WpAjPD/QZv/qzwm/tFvHVK3B862/SCA7T+UYL2da4CTJC1PVei/oU2aVwPn2v4TgKRzgNdQ3cW72/ZNZfktwCW2repO+PSy/U7A7pIOKa9b72hfbHvE6cRbvJbSPMf2jZJurLld054HPGT78XIn45XA14HXlTtVj1JdaG8q6X8AHAh8Hqo7VsN8L8O910LbA5L2pQoMR3I7sIGk6a7u5u7Zsu4i4AOSPlC+0y1tX18zH8vSaL/B5wHflLQR1YVy+ZZt6/7+RtrHSH4CfFFVjd45the2+V9cnqrmaQtgMbBxzX0PuovqO/wKcD7D1Or1iD+Vv6L6bvYeJt0T5e9ihp8sdArwKi+ppat2XB3/yTQx0Ei/3R/ZfhR4VNLDVEEvVOeilw23Q9t/kvRDYDdJt1EFEzcNk/wm4AuSPktVgL28XSJJHwP+bPs4SZsDmwMXl+9rKnB/zc+7rNzdcl6+FngRVQD/47Lsmzyzhvlpkp5XN20bz7q+t6z7rqua5lslPX+U/bwW+Ieyj/MlPdSy7pe2ryzPX0kVlPykfBcrUBXeB51T/l47uL8JdrXthQCqasmnUwUB99u+BsD2I2X9q4GvlGW3S/olS86vI/72S23L3wFntZyznzNK3s4sx//nku5iSTPl1uuMgP8n6bVUwdq6wPMZ+Xs9DLjK9mB/yk3o/v+PSaXfmjOJZ18kF1E+Z6n+WwHA9mVUJ5N7qWboe+cw+xvOEy3PB1peD7Dk4i7gLa6aGWxh+29s31bW/Ymx6cWL/4XAciXoOYqqSdO9VDU0VwH/S3Wn9eGS/oPAzFKteStVW++6/hPYV9KVVCfLEY9vKWD9M3ChpCuA37bk4yiqAseNqqrDjxpDPpal0X6DR1FdMDYH3kgVxA6q+/sbaR/Dsn008E9UTe+uHKY6/GCq4/5yqru8Kwyzu6f/h4sVy3s8VLa9lOou7mh9aHrBlVR9djYEkLSSpNGCq0ep7iwOGgzGKfvYojy9DHh7WbYLMGyTkz4x0m+3zvl7OCdS3WXdHzh5uES2/48ltdn/LunwoWkk7QDMZsm5TsAtLdeMl9reaZT8LGutx24xVS37stDu+j7oiSHpRjPcflrPi4MB/eB3sZntd7V5z5EC+k4aetyXY/hjsjRllynAH1s+9xa2XzxK3obmYfB16/F8O1Xtx1a2t6A69684wmeA6mbvVoPNrOiN/49Jpd+CiEuAf5S0BkD54d1DdSKHqpnH8mXd3wIP2P461VTfM0qap0rtBFQX3TeVC/nKVE1x2t5NGsbgHe3BtotbjvNztV78N2eEO2XdxPYTtnex/TLbs21vZ/tS4L9tb0zVxGAaVbtvbD9oe8+SfjPbwwYRto+w/YWW1z8v273S9qG2VynLL7W9W0u6A22fUl7+yPamVLVLK7bk48+231NOUJu3bt9jnkcVtEFV6KljaIF0PPtA0ots32T7s1THddNh9n1/uYO1D0tqj4amuwfYorSHXQ8Y7HO0JjDF9neo7ljNoMfZ/h3VcT69BN9X8szBB9r5H+DNpS3waxg+GP80Vdvx66hqSX81EZ+hi4zrtzsa21dR9e96G8/sS/UMqkaKedz2t6iaWs4Ysv5vqW5+/GNLrdEdwFqSXlXSLC/pJZ3K+wR5GHio/Pag9Htrl9B27bRttLu+j0fdYHo8Af2ydjvwQklbA6jqD7Ecz/yMG1O1gLijzg5LbcbdkmaX7SXp5aNsNrucn19E1Sep3Xs9j6rM9ZSqPhp/W5aP9L1eSNVH7nxV/epG+v8Yet2IZaCvmjPZvkXSZ4AfS1oMXE/VjvF7kq6m+rEORsbbAR+V9BTwGDBYEzGX6g70dbbfLukU4Oqy7kTb16saEaWOo4Avlf2JqjA0ngLp8cDJpVBxQ0t+etURqoZOXJHqrul3G8rHu0vTpxWofisnNJSPifI5quYcHwZ+WHObH7GkU/m/j3MfAAeVC8Viqtqm71Pd6VqkqqP9KVQFqO+Ui9WPWPK/eeOQdF8C7qa6o3szVadHqKrDT9aSUVQOHUP+ukZpTrd5y+sfAlu3STe95fl8qnPY4B3voTcW9hzyGtu/pwoeBvVyR/Q6xvvbreNMqn4BD42Q5qVUI7sNAE8B7xuyfj9gDeDccp/pPtu7quok+uXS9Gc5qt//LR3Of6ftC3xN0kpUzQz371Dapw1zfd9vHHn9NFWQfh1VANM2mLb9O0n7lbSDzXk+SdW3sSvYflJVh/2vqOpM/Geqm3P/SXWMb6Kqyd3P9hMavnn3UG8Hjpf0Saobr2cAC0ZIfwfVsXw+VR+Lv7R5r9OA/5E0n6occ3v5DCN+r7bPKgHEPKqRA4f7/zilfOY/06Y5Z0wM2b3YSiYmC0n7Ax8asvgntt/fRH4iIlTNOXCM7UuazktERFMSRERERNQgaXWqmuAFtmc3nZ+IiCYliGiYpJcCQ8dFfsL2K5rIT8RYpKYoJrvSlrtdjcQOpQnZpCTpOGDbIYuPtT1sZ/SyXY5nF5L0CapBAFqdZfszTeQnukOCiIiIiIiIGJN+G50pIiIiIiImWIKIiIiIiIgYkwQRERERERExJgkiIiIiIiJiTP4/e/povVjx2J4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 1080x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15, 4))\n",
    "sns.heatmap(customer_demographs.isnull(), cmap='viridis')  # Nulls "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Nulls in feature <i>**'no_of_childres'**</i> & <i>***'marital_status'***</i> "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Imputation - 'no_of_children' & 'marital_status'**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 760 entries, 0 to 759\n",
      "Data columns (total 7 columns):\n",
      "customer_id       760 non-null int64\n",
      "age_range         760 non-null object\n",
      "marital_status    431 non-null object\n",
      "rented            760 non-null int64\n",
      "family_size       760 non-null object\n",
      "no_of_children    222 non-null object\n",
      "income_bracket    760 non-null int64\n",
      "dtypes: int64(3), object(4)\n",
      "memory usage: 41.7+ KB\n"
     ]
    }
   ],
   "source": [
    "customer_demographs.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Replacing the 'NaN' with string 'NaN' for analysing distributions :\n",
    "\n",
    "customer_demographs['no_of_children'] = customer_demographs['no_of_children'].apply(lambda x:'NaN' if str(x)=='nan' else x)\n",
    "customer_demographs['marital_status'] = customer_demographs['marital_status'].apply(lambda x:'NaN' if str(x)=='nan' else x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0xed5700bec8>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(20, 10))\n",
    "\n",
    "plt.subplot(2,2,1)\n",
    "sns.countplot(data=customer_demographs, x='no_of_children', hue='marital_status')\n",
    "\n",
    "plt.subplot(2,2,2)\n",
    "sns.countplot(data=customer_demographs, x='no_of_children', hue='family_size')\n",
    "\n",
    "plt.subplot(2,2,3)\n",
    "sns.countplot(data=customer_demographs, x='no_of_children', hue='rented')\n",
    "\n",
    "plt.subplot(2,2,4)\n",
    "sns.countplot(data=customer_demographs, x='no_of_children', hue='income_bracket')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the 2nd plot, we infer that as no_of_children increases, family size increases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# NaN in both the features :\n",
    "\n",
    "len(customer_demographs[(customer_demographs['no_of_children'].isnull()) & (customer_demographs['marital_status'].isnull())])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(customer_demographs[customer_demographs['no_of_children'].isnull()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(customer_demographs[customer_demographs['marital_status'].isnull()])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<font color='darkblue'><b><i>We shall impute these Nulls using ML model itself later as the no of data points are too less to impute by any logic </i></b></font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1428"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# unique customer_ids in train set:\n",
    "\n",
    "train.customer_id.nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "760"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# unique customer_ids in customer_demographs set:\n",
    "\n",
    "customer_demographs.customer_id.nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No of common customer_ids for data points : 703\n",
      "No of customer_ids present in train but not in customer_demographs : 725\n"
     ]
    }
   ],
   "source": [
    "# Check for no of 'customer_ids' in 'train' set to exist in the 'customer_demographs' set :\n",
    "\n",
    "no_of_common_cust_ids = 0\n",
    "\n",
    "for each_cust_id in train.customer_id.unique():\n",
    "    \n",
    "    if each_cust_id in customer_demographs.customer_id.unique():\n",
    "        no_of_common_cust_ids+=1\n",
    "\n",
    "print('No of common customer_ids for data points :', no_of_common_cust_ids)\n",
    "print('No of customer_ids present in train but not in customer_demographs :', train.customer_id.nunique() - no_of_common_cust_ids)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Replacing the 'NaN' with string 'NaN' for analysing distributions :\n",
    "\n",
    "customer_demographs['no_of_children'] = customer_demographs['no_of_children'].apply(lambda x:np.NaN if str(x)=='NaN' else x)\n",
    "customer_demographs['marital_status'] = customer_demographs['marital_status'].apply(lambda x:np.NaN if str(x)=='NaN' else x)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### *Item_Data :*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "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>item_id</th>\n",
       "      <th>brand</th>\n",
       "      <th>brand_type</th>\n",
       "      <th>category</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>Established</td>\n",
       "      <td>Grocery</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>Established</td>\n",
       "      <td>Miscellaneous</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>56</td>\n",
       "      <td>Local</td>\n",
       "      <td>Bakery</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>56</td>\n",
       "      <td>Local</td>\n",
       "      <td>Grocery</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>56</td>\n",
       "      <td>Local</td>\n",
       "      <td>Grocery</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   item_id  brand   brand_type       category\n",
       "0        1      1  Established        Grocery\n",
       "1        2      1  Established  Miscellaneous\n",
       "2        3     56        Local         Bakery\n",
       "3        4     56        Local        Grocery\n",
       "4        5     56        Local        Grocery"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "item_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(74066, 4)"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "item_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 74066 entries, 0 to 74065\n",
      "Data columns (total 4 columns):\n",
      "item_id       74066 non-null int64\n",
      "brand         74066 non-null int64\n",
      "brand_type    74066 non-null object\n",
      "category      74066 non-null object\n",
      "dtypes: int64(2), object(2)\n",
      "memory usage: 2.3+ MB\n"
     ]
    }
   ],
   "source": [
    "item_data.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "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 tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>brand_type</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>brand_type</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>Established</td>\n",
       "      <td>62842</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Local</td>\n",
       "      <td>11224</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            brand_type\n",
       "                 count\n",
       "brand_type            \n",
       "Established      62842\n",
       "Local            11224"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "item_data.groupby('brand_type').agg({'brand_type':['count']})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "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>brand_type</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>category</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>Alcohol</td>\n",
       "      <td>303</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Bakery</td>\n",
       "      <td>1679</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Dairy, Juices &amp; Snacks</td>\n",
       "      <td>2425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Flowers &amp; Plants</td>\n",
       "      <td>664</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Fuel</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Garden</td>\n",
       "      <td>102</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Grocery</td>\n",
       "      <td>32448</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Meat</td>\n",
       "      <td>2080</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Miscellaneous</td>\n",
       "      <td>385</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Natural Products</td>\n",
       "      <td>2533</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Packaged Meat</td>\n",
       "      <td>1966</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Pharmaceutical</td>\n",
       "      <td>24471</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Prepared Food</td>\n",
       "      <td>1880</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Restauarant</td>\n",
       "      <td>78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Salads</td>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Seafood</td>\n",
       "      <td>728</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Skin &amp; Hair Care</td>\n",
       "      <td>2244</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Travel</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>Vegetables (cut)</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        brand_type\n",
       "category                          \n",
       "Alcohol                        303\n",
       "Bakery                        1679\n",
       "Dairy, Juices & Snacks        2425\n",
       "Flowers & Plants               664\n",
       "Fuel                            14\n",
       "Garden                         102\n",
       "Grocery                      32448\n",
       "Meat                          2080\n",
       "Miscellaneous                  385\n",
       "Natural Products              2533\n",
       "Packaged Meat                 1966\n",
       "Pharmaceutical               24471\n",
       "Prepared Food                 1880\n",
       "Restauarant                     78\n",
       "Salads                          40\n",
       "Seafood                        728\n",
       "Skin & Hair Care              2244\n",
       "Travel                          19\n",
       "Vegetables (cut)                 7"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "item_data.groupby('category').agg({'brand_type':'count'})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### *Coupon_item_mapping & Item_Data :*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "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>coupon_id</th>\n",
       "      <th>item_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>105</td>\n",
       "      <td>37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>107</td>\n",
       "      <td>75</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>494</td>\n",
       "      <td>76</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>522</td>\n",
       "      <td>77</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>518</td>\n",
       "      <td>77</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   coupon_id  item_id\n",
       "0        105       37\n",
       "1        107       75\n",
       "2        494       76\n",
       "3        522       77\n",
       "4        518       77"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "coupon_item_mapping.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "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>item_id</th>\n",
       "      <th>brand</th>\n",
       "      <th>brand_type</th>\n",
       "      <th>category</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>Established</td>\n",
       "      <td>Grocery</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>Established</td>\n",
       "      <td>Miscellaneous</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>56</td>\n",
       "      <td>Local</td>\n",
       "      <td>Bakery</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>56</td>\n",
       "      <td>Local</td>\n",
       "      <td>Grocery</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>56</td>\n",
       "      <td>Local</td>\n",
       "      <td>Grocery</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   item_id  brand   brand_type       category\n",
       "0        1      1  Established        Grocery\n",
       "1        2      1  Established  Miscellaneous\n",
       "2        3     56        Local         Bakery\n",
       "3        4     56        Local        Grocery\n",
       "4        5     56        Local        Grocery"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "item_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "coup_MERGED_item = coupon_item_mapping.merge(item_data, how='inner', on='item_id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "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>coupon_id</th>\n",
       "      <th>item_id</th>\n",
       "      <th>brand</th>\n",
       "      <th>brand_type</th>\n",
       "      <th>category</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>105</td>\n",
       "      <td>37</td>\n",
       "      <td>56</td>\n",
       "      <td>Local</td>\n",
       "      <td>Grocery</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>37</td>\n",
       "      <td>56</td>\n",
       "      <td>Local</td>\n",
       "      <td>Grocery</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>22</td>\n",
       "      <td>37</td>\n",
       "      <td>56</td>\n",
       "      <td>Local</td>\n",
       "      <td>Grocery</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>31</td>\n",
       "      <td>37</td>\n",
       "      <td>56</td>\n",
       "      <td>Local</td>\n",
       "      <td>Grocery</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>107</td>\n",
       "      <td>75</td>\n",
       "      <td>56</td>\n",
       "      <td>Local</td>\n",
       "      <td>Grocery</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   coupon_id  item_id  brand brand_type category\n",
       "0        105       37     56      Local  Grocery\n",
       "1          6       37     56      Local  Grocery\n",
       "2         22       37     56      Local  Grocery\n",
       "3         31       37     56      Local  Grocery\n",
       "4        107       75     56      Local  Grocery"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "coup_MERGED_item.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Established    78759\n",
       "Local          13904\n",
       "Name: brand_type, dtype: int64"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "coup_MERGED_item.brand_type.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Grocery                   36466\n",
       "Pharmaceutical            25061\n",
       "Natural Products           6819\n",
       "Meat                       6218\n",
       "Packaged Meat              6144\n",
       "Skin & Hair Care           4924\n",
       "Seafood                    2227\n",
       "Flowers & Plants           1963\n",
       "Dairy, Juices & Snacks     1867\n",
       "Garden                      286\n",
       "Prepared Food               240\n",
       "Miscellaneous               184\n",
       "Bakery                      100\n",
       "Salads                      100\n",
       "Travel                       44\n",
       "Vegetables (cut)             19\n",
       "Restauarant                   1\n",
       "Name: category, dtype: int64"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "coup_MERGED_item.category.value_counts()"
   ]
  },
  {
   "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>coupon_id</th>\n",
       "      <th>category</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>39</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>24</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   coupon_id  category\n",
       "0          1        39\n",
       "1          2         2\n",
       "2          3        17\n",
       "3          4        24\n",
       "4          5         7"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "coup_MERGED_item_groupBy = coup_MERGED_item.groupby('coupon_id').agg({'category':'count'})\n",
    "coup_MERGED_item_groupBy.reset_index(inplace=True)\n",
    "coup_MERGED_item_groupBy.head()  # The 'coupon_id' being applied to how many 'category'of products"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">   One coupon (coupon_id) gets applied on more than one type of item_id and each item_id can have different **(brand_type+category)**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### *Preparing the train set :* "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(78369, 8)"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# train_latest = train + campaign_data :\n",
    "\n",
    "train_MERGED_campaign_data = train.merge(campaign_data, how='left', on='campaign_id')\n",
    "train_MERGED_campaign_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "id                   0\n",
       "campaign_id          0\n",
       "coupon_id            0\n",
       "customer_id          0\n",
       "redemption_status    0\n",
       "campaign_type        0\n",
       "start_date           0\n",
       "end_date             0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_MERGED_campaign_data.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(78369, 9)"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# train_latest = train_latest + coup_merged_item\n",
    "train_MERGED_campaign_data_MERGED_coupon = train_MERGED_campaign_data.merge(coup_MERGED_item_groupBy, how='left', on='coupon_id') \n",
    "train_MERGED_campaign_data_MERGED_coupon.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "id                   0\n",
       "campaign_id          0\n",
       "coupon_id            0\n",
       "customer_id          0\n",
       "redemption_status    0\n",
       "campaign_type        0\n",
       "start_date           0\n",
       "end_date             0\n",
       "category             0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_MERGED_campaign_data_MERGED_coupon.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(78369, 15)"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# train_latest = train_latest + customer_dempgraphs\n",
    "train_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics = train_MERGED_campaign_data_MERGED_coupon.merge(customer_demographs, \n",
    "                                                                                                                      how='left', on='customer_id')\n",
    "train_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics.shape                                                                        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "id                       0\n",
       "campaign_id              0\n",
       "coupon_id                0\n",
       "customer_id              0\n",
       "redemption_status        0\n",
       "campaign_type            0\n",
       "start_date               0\n",
       "end_date                 0\n",
       "category                 0\n",
       "age_range            34708\n",
       "marital_status       52975\n",
       "rented               34708\n",
       "family_size          34708\n",
       "no_of_children       64313\n",
       "income_bracket       34708\n",
       "dtype: int64"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(78369, 27)"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# train_latest = train_latest + customer_transaction_data_merged_item_data\n",
    "train_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics_MERGED_customer_transaction_data = train_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics.merge(customer_transaction_data_groupBy, \n",
    "                                                                                                                      how='left', \n",
    "                                                                                                                      on='customer_id')\n",
    "train_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics_MERGED_customer_transaction_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "id                             0\n",
       "campaign_id                    0\n",
       "coupon_id                      0\n",
       "customer_id                    0\n",
       "redemption_status              0\n",
       "campaign_type                  0\n",
       "start_date                     0\n",
       "end_date                       0\n",
       "category                       0\n",
       "age_range                  34708\n",
       "marital_status             52975\n",
       "rented                     34708\n",
       "family_size                34708\n",
       "no_of_children             64313\n",
       "income_bracket             34708\n",
       "(selling_price, mean)          0\n",
       "(selling_price, max)           0\n",
       "(selling_price, min)           0\n",
       "(selling_price, sum)           0\n",
       "(other_discount, mean)         0\n",
       "(other_discount, max)          0\n",
       "(other_discount, min)          0\n",
       "(other_discount, sum)          0\n",
       "(coupon_discount, mean)        0\n",
       "(coupon_discount, max)         0\n",
       "(coupon_discount, min)         0\n",
       "(coupon_discount, sum)         0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics_MERGED_customer_transaction_data.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(78369, 27)"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# final train_set :\n",
    "\n",
    "train_df = train_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics_MERGED_customer_transaction_data\n",
    "train_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "id                             0\n",
       "campaign_id                    0\n",
       "coupon_id                      0\n",
       "customer_id                    0\n",
       "redemption_status              0\n",
       "campaign_type                  0\n",
       "start_date                     0\n",
       "end_date                       0\n",
       "category                       0\n",
       "age_range                  34708\n",
       "marital_status             52975\n",
       "rented                     34708\n",
       "family_size                34708\n",
       "no_of_children             64313\n",
       "income_bracket             34708\n",
       "(selling_price, mean)          0\n",
       "(selling_price, max)           0\n",
       "(selling_price, min)           0\n",
       "(selling_price, sum)           0\n",
       "(other_discount, mean)         0\n",
       "(other_discount, max)          0\n",
       "(other_discount, min)          0\n",
       "(other_discount, sum)          0\n",
       "(coupon_discount, mean)        0\n",
       "(coupon_discount, max)         0\n",
       "(coupon_discount, min)         0\n",
       "(coupon_discount, sum)         0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Rectifying columns in train :\n",
    "feature_list_train = [str(each_train_col[0]+\"_\"+each_train_col[1]) if str(type(each_train_col))==\"<class 'tuple'>\" else each_train_col  for each_train_col in train_df.columns ]\n",
    "\n",
    "# rectifying column names :\n",
    "train_df.columns = feature_list_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "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>id</th>\n",
       "      <th>campaign_id</th>\n",
       "      <th>coupon_id</th>\n",
       "      <th>customer_id</th>\n",
       "      <th>redemption_status</th>\n",
       "      <th>campaign_type</th>\n",
       "      <th>start_date</th>\n",
       "      <th>end_date</th>\n",
       "      <th>category</th>\n",
       "      <th>age_range</th>\n",
       "      <th>...</th>\n",
       "      <th>selling_price_min</th>\n",
       "      <th>selling_price_sum</th>\n",
       "      <th>other_discount_mean</th>\n",
       "      <th>other_discount_max</th>\n",
       "      <th>other_discount_min</th>\n",
       "      <th>other_discount_sum</th>\n",
       "      <th>coupon_discount_mean</th>\n",
       "      <th>coupon_discount_max</th>\n",
       "      <th>coupon_discount_min</th>\n",
       "      <th>coupon_discount_sum</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>13</td>\n",
       "      <td>27</td>\n",
       "      <td>1053</td>\n",
       "      <td>0</td>\n",
       "      <td>X</td>\n",
       "      <td>2013-05-19</td>\n",
       "      <td>2013-05-07</td>\n",
       "      <td>125</td>\n",
       "      <td>46-55</td>\n",
       "      <td>...</td>\n",
       "      <td>17.45</td>\n",
       "      <td>57120.75</td>\n",
       "      <td>-33.168935</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-890.50</td>\n",
       "      <td>-10282.37</td>\n",
       "      <td>-0.287258</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-89.05</td>\n",
       "      <td>-89.05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>13</td>\n",
       "      <td>116</td>\n",
       "      <td>48</td>\n",
       "      <td>0</td>\n",
       "      <td>X</td>\n",
       "      <td>2013-05-19</td>\n",
       "      <td>2013-05-07</td>\n",
       "      <td>3</td>\n",
       "      <td>36-45</td>\n",
       "      <td>...</td>\n",
       "      <td>7.12</td>\n",
       "      <td>90185.10</td>\n",
       "      <td>-27.699169</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-372.23</td>\n",
       "      <td>-10664.18</td>\n",
       "      <td>-3.215039</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-213.72</td>\n",
       "      <td>-1237.79</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>9</td>\n",
       "      <td>635</td>\n",
       "      <td>205</td>\n",
       "      <td>0</td>\n",
       "      <td>Y</td>\n",
       "      <td>2013-11-03</td>\n",
       "      <td>2013-12-04</td>\n",
       "      <td>67</td>\n",
       "      <td>46-55</td>\n",
       "      <td>...</td>\n",
       "      <td>17.45</td>\n",
       "      <td>117461.66</td>\n",
       "      <td>-17.795660</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-468.05</td>\n",
       "      <td>-17261.79</td>\n",
       "      <td>-2.212082</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-89.05</td>\n",
       "      <td>-2145.72</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>13</td>\n",
       "      <td>644</td>\n",
       "      <td>1050</td>\n",
       "      <td>0</td>\n",
       "      <td>X</td>\n",
       "      <td>2013-05-19</td>\n",
       "      <td>2013-05-07</td>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>13.89</td>\n",
       "      <td>23291.42</td>\n",
       "      <td>-16.655570</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-312.03</td>\n",
       "      <td>-3947.37</td>\n",
       "      <td>-0.751477</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-89.05</td>\n",
       "      <td>-178.10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>9</td>\n",
       "      <td>8</td>\n",
       "      <td>1017</td>\n",
       "      <td>1489</td>\n",
       "      <td>0</td>\n",
       "      <td>X</td>\n",
       "      <td>2013-02-16</td>\n",
       "      <td>2013-05-04</td>\n",
       "      <td>32</td>\n",
       "      <td>46-55</td>\n",
       "      <td>...</td>\n",
       "      <td>8.90</td>\n",
       "      <td>67797.49</td>\n",
       "      <td>-20.524733</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-320.22</td>\n",
       "      <td>-11534.90</td>\n",
       "      <td>-0.471548</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-53.43</td>\n",
       "      <td>-265.01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 27 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   id  campaign_id  coupon_id  customer_id  redemption_status campaign_type  \\\n",
       "0   1           13         27         1053                  0             X   \n",
       "1   2           13        116           48                  0             X   \n",
       "2   6            9        635          205                  0             Y   \n",
       "3   7           13        644         1050                  0             X   \n",
       "4   9            8       1017         1489                  0             X   \n",
       "\n",
       "  start_date   end_date  category age_range  ... selling_price_min  \\\n",
       "0 2013-05-19 2013-05-07       125     46-55  ...             17.45   \n",
       "1 2013-05-19 2013-05-07         3     36-45  ...              7.12   \n",
       "2 2013-11-03 2013-12-04        67     46-55  ...             17.45   \n",
       "3 2013-05-19 2013-05-07         4       NaN  ...             13.89   \n",
       "4 2013-02-16 2013-05-04        32     46-55  ...              8.90   \n",
       "\n",
       "   selling_price_sum other_discount_mean other_discount_max  \\\n",
       "0           57120.75          -33.168935                0.0   \n",
       "1           90185.10          -27.699169                0.0   \n",
       "2          117461.66          -17.795660                0.0   \n",
       "3           23291.42          -16.655570                0.0   \n",
       "4           67797.49          -20.524733                0.0   \n",
       "\n",
       "   other_discount_min  other_discount_sum  coupon_discount_mean  \\\n",
       "0             -890.50           -10282.37             -0.287258   \n",
       "1             -372.23           -10664.18             -3.215039   \n",
       "2             -468.05           -17261.79             -2.212082   \n",
       "3             -312.03            -3947.37             -0.751477   \n",
       "4             -320.22           -11534.90             -0.471548   \n",
       "\n",
       "   coupon_discount_max  coupon_discount_min  coupon_discount_sum  \n",
       "0                  0.0               -89.05               -89.05  \n",
       "1                  0.0              -213.72             -1237.79  \n",
       "2                  0.0               -89.05             -2145.72  \n",
       "3                  0.0               -89.05              -178.10  \n",
       "4                  0.0               -53.43              -265.01  \n",
       "\n",
       "[5 rows x 27 columns]"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Feature engineering timedelta(days) from campaign_start and end_dates for train and test sets :\n",
    "\n",
    "train_df['Gap_days_campaign'] = train_df['end_date'] - train_df['start_date'] \n",
    "train_df['Gap_days_campaign'] = train_df['Gap_days_campaign'].apply(lambda timedelta:int(np.abs(timedelta.days)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### *Preparing the test set :* "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50226, 4)"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50226, 26)"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# test_latest = test + campaign_data \n",
    "\n",
    "test_MERGED_campaign_data = test.merge(campaign_data, how='left', on='campaign_id')\n",
    "test_MERGED_campaign_data.shape\n",
    "\n",
    "# test_latest = test_latest + coup_merged_item\n",
    "test_MERGED_campaign_data_MERGED_coupon = test_MERGED_campaign_data.merge(coup_MERGED_item_groupBy, how='left', on='coupon_id') \n",
    "# test_MERGED_campaign_data_MERGED_coupon.shape\n",
    "\n",
    "# test_latest = test_latest + customer_dempgraphs\n",
    "test_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics = test_MERGED_campaign_data_MERGED_coupon.merge(customer_demographs, \n",
    "                                                                                                                      how='left', on='customer_id')\n",
    "test_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics.shape                                                                        \n",
    "\n",
    "# test_latest = test_latest + customer_transaction_data_merged_item_data\n",
    "test_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics_MERGED_customer_transaction_data = test_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics.merge(customer_transaction_data_groupBy, \n",
    "                                                                                                                      how='left', \n",
    "                                                                                                                      on='customer_id')\n",
    "# test_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics_MERGED_customer_transaction_data.shape\n",
    "\n",
    "# final test_set :\n",
    "\n",
    "test_df = test_MERGED_campaign_data_MERGED_coupon_MERGED_customer_demographics_MERGED_customer_transaction_data\n",
    "test_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Feature engineering timedelta(days) from campaign_start and end_dates for train and test sets :\n",
    "\n",
    "test_df['Gap_days_campaign'] = test_df['end_date'] - test_df['start_date'] \n",
    "test_df['Gap_days_campaign'] = test_df['Gap_days_campaign'].apply(lambda timedelta:int(np.abs(timedelta.days)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Rectifying columns in test :\n",
    "\n",
    "feature_list_test = [str(each_test_col[0]+\"_\"+each_test_col[1]) if str(type(each_test_col))==\"<class 'tuple'>\" else each_test_col for each_test_col in test_df.columns]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Assigning proper columns names :\n",
    "\n",
    "test_df.columns = feature_list_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Eliminating irrelvant features in the test set :\n",
    "\n",
    "del test_df['id']\n",
    "del test_df['campaign_id']\n",
    "del test_df['customer_id']\n",
    "del test_df['coupon_id']\n",
    "del test_df['start_date']\n",
    "del test_df['end_date']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "campaign_type               0\n",
       "category                    0\n",
       "age_range               19287\n",
       "marital_status          32144\n",
       "rented                  19287\n",
       "family_size             19287\n",
       "no_of_children          40064\n",
       "income_bracket          19287\n",
       "selling_price_mean          0\n",
       "selling_price_max           0\n",
       "selling_price_min           0\n",
       "selling_price_sum           0\n",
       "other_discount_mean         0\n",
       "other_discount_max          0\n",
       "other_discount_min          0\n",
       "other_discount_sum          0\n",
       "coupon_discount_mean        0\n",
       "coupon_discount_max         0\n",
       "coupon_discount_min         0\n",
       "coupon_discount_sum         0\n",
       "Gap_days_campaign           0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_df.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <font color='red'>Feature imputation using Modelling Approach :</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0xed576dc208>"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x504 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(20, 7))\n",
    "sns.heatmap(train_df.isnull(), cmap=\"YlGnBu\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "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>id</th>\n",
       "      <th>campaign_id</th>\n",
       "      <th>coupon_id</th>\n",
       "      <th>customer_id</th>\n",
       "      <th>redemption_status</th>\n",
       "      <th>campaign_type</th>\n",
       "      <th>start_date</th>\n",
       "      <th>end_date</th>\n",
       "      <th>category</th>\n",
       "      <th>age_range</th>\n",
       "      <th>...</th>\n",
       "      <th>selling_price_sum</th>\n",
       "      <th>other_discount_mean</th>\n",
       "      <th>other_discount_max</th>\n",
       "      <th>other_discount_min</th>\n",
       "      <th>other_discount_sum</th>\n",
       "      <th>coupon_discount_mean</th>\n",
       "      <th>coupon_discount_max</th>\n",
       "      <th>coupon_discount_min</th>\n",
       "      <th>coupon_discount_sum</th>\n",
       "      <th>Gap_days_campaign</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>13</td>\n",
       "      <td>27</td>\n",
       "      <td>1053</td>\n",
       "      <td>0</td>\n",
       "      <td>X</td>\n",
       "      <td>2013-05-19</td>\n",
       "      <td>2013-05-07</td>\n",
       "      <td>125</td>\n",
       "      <td>46-55</td>\n",
       "      <td>...</td>\n",
       "      <td>57120.75</td>\n",
       "      <td>-33.168935</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-890.50</td>\n",
       "      <td>-10282.37</td>\n",
       "      <td>-0.287258</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-89.05</td>\n",
       "      <td>-89.05</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>13</td>\n",
       "      <td>116</td>\n",
       "      <td>48</td>\n",
       "      <td>0</td>\n",
       "      <td>X</td>\n",
       "      <td>2013-05-19</td>\n",
       "      <td>2013-05-07</td>\n",
       "      <td>3</td>\n",
       "      <td>36-45</td>\n",
       "      <td>...</td>\n",
       "      <td>90185.10</td>\n",
       "      <td>-27.699169</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-372.23</td>\n",
       "      <td>-10664.18</td>\n",
       "      <td>-3.215039</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-213.72</td>\n",
       "      <td>-1237.79</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>9</td>\n",
       "      <td>635</td>\n",
       "      <td>205</td>\n",
       "      <td>0</td>\n",
       "      <td>Y</td>\n",
       "      <td>2013-11-03</td>\n",
       "      <td>2013-12-04</td>\n",
       "      <td>67</td>\n",
       "      <td>46-55</td>\n",
       "      <td>...</td>\n",
       "      <td>117461.66</td>\n",
       "      <td>-17.795660</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-468.05</td>\n",
       "      <td>-17261.79</td>\n",
       "      <td>-2.212082</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-89.05</td>\n",
       "      <td>-2145.72</td>\n",
       "      <td>31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>13</td>\n",
       "      <td>644</td>\n",
       "      <td>1050</td>\n",
       "      <td>0</td>\n",
       "      <td>X</td>\n",
       "      <td>2013-05-19</td>\n",
       "      <td>2013-05-07</td>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>23291.42</td>\n",
       "      <td>-16.655570</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-312.03</td>\n",
       "      <td>-3947.37</td>\n",
       "      <td>-0.751477</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-89.05</td>\n",
       "      <td>-178.10</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>9</td>\n",
       "      <td>8</td>\n",
       "      <td>1017</td>\n",
       "      <td>1489</td>\n",
       "      <td>0</td>\n",
       "      <td>X</td>\n",
       "      <td>2013-02-16</td>\n",
       "      <td>2013-05-04</td>\n",
       "      <td>32</td>\n",
       "      <td>46-55</td>\n",
       "      <td>...</td>\n",
       "      <td>67797.49</td>\n",
       "      <td>-20.524733</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-320.22</td>\n",
       "      <td>-11534.90</td>\n",
       "      <td>-0.471548</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-53.43</td>\n",
       "      <td>-265.01</td>\n",
       "      <td>77</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 28 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   id  campaign_id  coupon_id  customer_id  redemption_status campaign_type  \\\n",
       "0   1           13         27         1053                  0             X   \n",
       "1   2           13        116           48                  0             X   \n",
       "2   6            9        635          205                  0             Y   \n",
       "3   7           13        644         1050                  0             X   \n",
       "4   9            8       1017         1489                  0             X   \n",
       "\n",
       "  start_date   end_date  category age_range  ... selling_price_sum  \\\n",
       "0 2013-05-19 2013-05-07       125     46-55  ...          57120.75   \n",
       "1 2013-05-19 2013-05-07         3     36-45  ...          90185.10   \n",
       "2 2013-11-03 2013-12-04        67     46-55  ...         117461.66   \n",
       "3 2013-05-19 2013-05-07         4       NaN  ...          23291.42   \n",
       "4 2013-02-16 2013-05-04        32     46-55  ...          67797.49   \n",
       "\n",
       "   other_discount_mean other_discount_max other_discount_min  \\\n",
       "0           -33.168935                0.0            -890.50   \n",
       "1           -27.699169                0.0            -372.23   \n",
       "2           -17.795660                0.0            -468.05   \n",
       "3           -16.655570                0.0            -312.03   \n",
       "4           -20.524733                0.0            -320.22   \n",
       "\n",
       "   other_discount_sum  coupon_discount_mean  coupon_discount_max  \\\n",
       "0           -10282.37             -0.287258                  0.0   \n",
       "1           -10664.18             -3.215039                  0.0   \n",
       "2           -17261.79             -2.212082                  0.0   \n",
       "3            -3947.37             -0.751477                  0.0   \n",
       "4           -11534.90             -0.471548                  0.0   \n",
       "\n",
       "   coupon_discount_min  coupon_discount_sum  Gap_days_campaign  \n",
       "0               -89.05               -89.05                 12  \n",
       "1              -213.72             -1237.79                 12  \n",
       "2               -89.05             -2145.72                 31  \n",
       "3               -89.05              -178.10                 12  \n",
       "4               -53.43              -265.01                 77  \n",
       "\n",
       "[5 rows x 28 columns]"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "id                          0\n",
       "campaign_id                 0\n",
       "coupon_id                   0\n",
       "customer_id                 0\n",
       "redemption_status           0\n",
       "campaign_type               0\n",
       "start_date                  0\n",
       "end_date                    0\n",
       "category                    0\n",
       "age_range               34708\n",
       "marital_status          52975\n",
       "rented                  34708\n",
       "family_size             34708\n",
       "no_of_children          64313\n",
       "income_bracket          34708\n",
       "selling_price_mean          0\n",
       "selling_price_max           0\n",
       "selling_price_min           0\n",
       "selling_price_sum           0\n",
       "other_discount_mean         0\n",
       "other_discount_max          0\n",
       "other_discount_min          0\n",
       "other_discount_sum          0\n",
       "coupon_discount_mean        0\n",
       "coupon_discount_max         0\n",
       "coupon_discount_min         0\n",
       "coupon_discount_sum         0\n",
       "Gap_days_campaign           0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0xed592ce408>"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Excluding nulls :\n",
    "\n",
    "plt.figure(figsize=(20, 10))\n",
    "\n",
    "plt.subplot(2,2,1)\n",
    "sns.countplot(data=train_df, x='no_of_children', hue='marital_status')\n",
    "\n",
    "plt.subplot(2,2,2)\n",
    "sns.countplot(data=train_df, x='no_of_children', hue='family_size')\n",
    "\n",
    "plt.subplot(2,2,3)\n",
    "sns.countplot(data=train_df, x='no_of_children', hue='rented')\n",
    "\n",
    "plt.subplot(2,2,4)\n",
    "sns.countplot(data=train_df, x='no_of_children', hue='income_bracket')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h3> Train set features to impute :</h3>\n",
    "\n",
    "+ age_range               \n",
    "+ marital_status          \n",
    "+ rented                  \n",
    "+ family_size             \n",
    "+ no_of_children          \n",
    "+ income_bracket          "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "# removing irrelevant features from the train set:\n",
    "\n",
    "del train_df['id']\n",
    "del train_df['campaign_id']\n",
    "del train_df['customer_id']\n",
    "del train_df['coupon_id']\n",
    "del train_df['start_date']\n",
    "del train_df['end_date']\n",
    "\n",
    "# redemption_status= train_df['redemption_status'] # storing the 'y' dependent variable, in separete variable :\n",
    "# del train_df['redemption_status']  # deleting the dependent variale"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----------------+-------------------+-----------+----------------+--------+-------------+----------------+----------------+\n",
      "| FeatureToImpute | redemption_status | age_range | marital_status | rented | family_size | no_of_children | income_Bracket |\n",
      "+-----------------+-------------------+-----------+----------------+--------+-------------+----------------+----------------+\n",
      "|    age_range    |        N/A        |    N/A    |       X        |   X    |      X      |       X        |       X        |\n",
      "|  marital_status |        N/A        |     ✓     |      N/A       |   X    |      X      |       X        |       X        |\n",
      "|      rented     |        N/A        |     ✓     |       ✓        |  N/A   |      X      |       X        |       X        |\n",
      "|   family_size   |        N/A        |     ✓     |       ✓        |   ✓    |     N/A     |       X        |       X        |\n",
      "|  no_of_children |        N/A        |     ✓     |       ✓        |   ✓    |      ✓      |      N/A       |       X        |\n",
      "|  income_bracket |        N/A        |     ✓     |       ✓        |   ✓    |      ✓      |       ✓        |      N/A       |\n",
      "+-----------------+-------------------+-----------+----------------+--------+-------------+----------------+----------------+\n"
     ]
    }
   ],
   "source": [
    "# Utilization of the features in incremental way to achieve imputation in the 'after' features : \n",
    "# The models will be used exactly in the same step as they were used for imputation in the train set :\n",
    "\n",
    "table = PrettyTable()\n",
    "table.field_names = ['FeatureToImpute', 'redemption_status','age_range', 'marital_status', 'rented', 'family_size', 'no_of_children', 'income_Bracket']\n",
    "\n",
    "table.align[\"FeatureToImpute \"] = \"L\"\n",
    "table.align[\"redemption_status\"] = \"L\"\n",
    "table.align[\"age_range\"] = \"R\"\n",
    "table.align[\"marital_status \"] = \"R\"\n",
    "table.align[\"rented\"] = \"R\"\n",
    "table.align[\"family_size\"] = \"R\"\n",
    "table.align[\"no_of_children\"] = \"R\"\n",
    "table.align[\"income_bracket\"] = \"L\"\n",
    "\n",
    "\n",
    "table.add_row(['age_range','N/A', 'N/A', 'X' , 'X' , 'X' , 'X' , 'X'  ])\n",
    "\n",
    "table.add_row(['marital_status','N/A', '✓', 'N/A', 'X' , 'X' , 'X' , 'X' ])\n",
    "\n",
    "table.add_row(['rented','N/A', '✓', '✓', 'N/A', 'X' , 'X' , 'X' ])\n",
    "\n",
    "table.add_row(['family_size','N/A', '✓', '✓', '✓', 'N/A', 'X' , 'X' ])\n",
    "\n",
    "table.add_row(['no_of_children', 'N/A', '✓', '✓', '✓', '✓', 'N/A', 'X' ])\n",
    "\n",
    "table.add_row(['income_bracket', 'N/A', '✓', '✓', '✓', '✓', '✓', 'N/A'])\n",
    "\n",
    "print(table)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='red'> 'age_range' </font> for the train set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(78369, 23)"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# One hot encoding 'campaign_type' feature :\n",
    "\n",
    "train_df = pd.concat([train_df, pd.get_dummies(train_df['campaign_type'])], axis=1)\n",
    "del train_df['campaign_type'] # has been one hot encoded, hence not required as a single feature\n",
    "train_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'age_range' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_age_range = train_df.ix[train_df['age_range'].notnull()] # where age_range is notNull\n",
    "\n",
    "# The following features are not considered for as a training set feature for 'age_range' imputation becoz they themselves contain Nulls \n",
    "\n",
    "\n",
    "# Redemption_status not an inpput feature as it will not be present in the test set and,\n",
    "# the models used for imuting in train set will be used for test set imputation as well \n",
    "\n",
    "X_train_redemption_status = X_train_impute_age_range['redemption_status']\n",
    "del X_train_impute_age_range['redemption_status']\n",
    "\n",
    "X_train_marital_status = X_train_impute_age_range['marital_status']\n",
    "del X_train_impute_age_range['marital_status']\n",
    "\n",
    "X_train_rented = X_train_impute_age_range['rented']\n",
    "del X_train_impute_age_range['rented']\n",
    "\n",
    "X_train_family_size = X_train_impute_age_range['family_size']\n",
    "del X_train_impute_age_range['family_size']\n",
    "\n",
    "X_train_no_of_children = X_train_impute_age_range['no_of_children']\n",
    "del X_train_impute_age_range['no_of_children']\n",
    "\n",
    "X_train_income_bracket = X_train_impute_age_range['income_bracket'] \n",
    "del X_train_impute_age_range['income_bracket']\n",
    "\n",
    "\n",
    "y_train_age_range = X_train_impute_age_range['age_range']  # y variable i.e. age_range(with all notNull values)\n",
    "del X_train_impute_age_range['age_range']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(43661, 16)"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_train_impute_age_range shape :\n",
    "\n",
    "X_train_impute_age_range.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "category                0\n",
       "selling_price_mean      0\n",
       "selling_price_max       0\n",
       "selling_price_min       0\n",
       "selling_price_sum       0\n",
       "other_discount_mean     0\n",
       "other_discount_max      0\n",
       "other_discount_min      0\n",
       "other_discount_sum      0\n",
       "coupon_discount_mean    0\n",
       "coupon_discount_max     0\n",
       "coupon_discount_min     0\n",
       "coupon_discount_sum     0\n",
       "Gap_days_campaign       0\n",
       "X                       0\n",
       "Y                       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_age_range.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "                   intercept_scaling=1, l1_ratio=None, max_iter=100,\n",
       "                   multi_class='warn', n_jobs=None, penalty='l2',\n",
       "                   random_state=None, solver='warn', tol=0.0001, verbose=0,\n",
       "                   warm_start=False)"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Imputation using Model for 'age_range' features :\n",
    "\n",
    "logisticRegression_age_range = LogisticRegression()\n",
    "logisticRegression_age_range.fit(X_train_impute_age_range, y_train_age_range)  # LR fitted with X, y "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_age_range contains\n",
    "\n",
    "X_test_impute_age_range = train_df.ix[train_df['age_range'].isnull()] # all the points containing 'age_range' Nulls\n",
    "\n",
    "\n",
    "# Redemption_status not an inpput feature as it will not be present in the test set and,\n",
    "# the models used for imuting in train set will be used for test set imputation as well \n",
    "\n",
    "X_test_redemption_status = X_test_impute_age_range['redemption_status']\n",
    "del X_test_impute_age_range['redemption_status']\n",
    "\n",
    "\n",
    "\n",
    "X_test_marital_status = X_test_impute_age_range['marital_status']\n",
    "del X_test_impute_age_range['marital_status']\n",
    "\n",
    "X_test_rented = X_test_impute_age_range['rented']\n",
    "del X_test_impute_age_range['rented']\n",
    "\n",
    "X_test_family_size = X_test_impute_age_range['family_size']\n",
    "del X_test_impute_age_range['family_size']\n",
    "\n",
    "X_test_no_of_children = X_test_impute_age_range['no_of_children']\n",
    "del X_test_impute_age_range['no_of_children']\n",
    "\n",
    "X_test_income_bracket = X_test_impute_age_range['income_bracket'] \n",
    "del X_test_impute_age_range['income_bracket']\n",
    "\n",
    "# Contains all Nulls; this step not important though at all as this 'y' will have new not null values \n",
    "y_test_age_range = X_test_impute_age_range['age_range']  \n",
    "del X_test_impute_age_range['age_range']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(34708, 16)"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_age_range.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(43661, 16)"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_age_range.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "category                0\n",
       "selling_price_mean      0\n",
       "selling_price_max       0\n",
       "selling_price_min       0\n",
       "selling_price_sum       0\n",
       "other_discount_mean     0\n",
       "other_discount_max      0\n",
       "other_discount_min      0\n",
       "other_discount_sum      0\n",
       "coupon_discount_mean    0\n",
       "coupon_discount_max     0\n",
       "coupon_discount_min     0\n",
       "coupon_discount_sum     0\n",
       "Gap_days_campaign       0\n",
       "X                       0\n",
       "Y                       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_age_range.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_age_range = logisticRegression_age_range.predict(X_test_impute_age_range)\n",
    "X_test_impute_age_range['age_range'] = predicted_age_range # imputation via predicted values for the nulls"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'age_range' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "\n",
    "# Completing the X_train set :\n",
    "X_train_impute_age_range['age_range'] = y_train_age_range\n",
    "X_train_impute_age_range['marital_status'] = X_train_marital_status\n",
    "X_train_impute_age_range['rented'] = X_train_rented\n",
    "X_train_impute_age_range['family_size'] = X_train_family_size\n",
    "X_train_impute_age_range['no_of_children'] = X_train_no_of_children\n",
    "X_train_impute_age_range['income_bracket'] = X_train_income_bracket\n",
    "X_train_impute_age_range['redemption_status'] = X_train_redemption_status\n",
    "\n",
    "\n",
    "# Completing the X_test set :\n",
    "# X_test_impute_age_range['age_range'] = predicted_age_range # imputation via predicted values for the nulls # already imputed in above cell\n",
    "X_test_impute_age_range['marital_status'] = X_test_marital_status \n",
    "X_test_impute_age_range['rented'] = X_test_rented\n",
    "X_test_impute_age_range['family_size'] = X_test_family_size\n",
    "X_test_impute_age_range['no_of_children'] = X_test_no_of_children\n",
    "X_test_impute_age_range['income_bracket'] = X_test_income_bracket\n",
    "X_test_impute_age_range['redemption_status'] = X_test_redemption_status"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (43661, 23)\n",
      "X_test shape : (34708, 23)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_age_range.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_age_range.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train_df shape : (78369, 23)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "train_df = X_train_impute_age_range.append(X_test_impute_age_range)\n",
    "print('Train_df shape :', train_df.shape)\n",
    "\n",
    "# 'age_range' feature IMPUTED *******"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='red'> 'marital_status' </font> for the train set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(78369, 28)"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# One hot encoding 'campaign_type' feature :\n",
    "\n",
    "train_df = pd.concat([train_df, pd.get_dummies(train_df['age_range'])], axis=1)\n",
    "del train_df['age_range'] # has been one hot encoded, hence not required as a single feature\n",
    "train_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'age_range' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_marital_status = train_df.ix[train_df['marital_status'].notnull()] # where age_range is notNull\n",
    "\n",
    "# Redemption_status not an inpput feature as it will not be present in the test set and,\n",
    "# the models used for imuting in train set will be used for test set imputation as well \n",
    "\n",
    "X_train_redemption_status = X_train_impute_marital_status['redemption_status']\n",
    "del X_train_impute_marital_status['redemption_status']\n",
    "\n",
    "\n",
    "# The following features are not considered for as a training set feature for 'age_range' imputation becoz they themselves contain Nulls \n",
    "\n",
    "X_train_rented = X_train_impute_marital_status['rented']\n",
    "del X_train_impute_marital_status['rented']\n",
    "\n",
    "X_train_family_size = X_train_impute_marital_status['family_size']\n",
    "del X_train_impute_marital_status['family_size']\n",
    "\n",
    "X_train_no_of_children = X_train_impute_marital_status['no_of_children']\n",
    "del X_train_impute_marital_status['no_of_children']\n",
    "\n",
    "X_train_income_bracket = X_train_impute_marital_status['income_bracket'] \n",
    "del X_train_impute_marital_status['income_bracket']\n",
    "\n",
    "\n",
    "y_train_marital_status = X_train_impute_marital_status['marital_status']  # y variable i.e. age_range(with all notNull values)\n",
    "del X_train_impute_marital_status['marital_status']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25394, 22)"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_train_impute_age_range shape :\n",
    "\n",
    "X_train_impute_marital_status.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "                   intercept_scaling=1, l1_ratio=None, max_iter=100,\n",
       "                   multi_class='warn', n_jobs=None, penalty='l2',\n",
       "                   random_state=None, solver='warn', tol=0.0001, verbose=0,\n",
       "                   warm_start=False)"
      ]
     },
     "execution_count": 86,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Imputation using Model for 'age_range' features :\n",
    "\n",
    "logisticRegression_marital_status = LogisticRegression()\n",
    "logisticRegression_marital_status.fit(X_train_impute_marital_status, y_train_marital_status)  # LR fitted with X, y "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_marital_status contains\n",
    "\n",
    "X_test_impute_marital_status = train_df.ix[train_df['marital_status'].isnull()] # all the points containing 'marial_status' Nulls\n",
    "\n",
    "X_test_redemption_status = X_test_impute_marital_status['redemption_status']\n",
    "del X_test_impute_marital_status['redemption_status']\n",
    "\n",
    "\n",
    "X_test_rented = X_test_impute_marital_status['rented']\n",
    "del X_test_impute_marital_status['rented']\n",
    "\n",
    "X_test_family_size = X_test_impute_marital_status['family_size']\n",
    "del X_test_impute_marital_status['family_size']\n",
    "\n",
    "X_test_no_of_children = X_test_impute_marital_status['no_of_children']\n",
    "del X_test_impute_marital_status['no_of_children']\n",
    "\n",
    "X_test_income_bracket = X_test_impute_marital_status['income_bracket'] \n",
    "del X_test_impute_marital_status['income_bracket']\n",
    "\n",
    "\n",
    "y_test_marital_status = X_test_impute_marital_status['marital_status']  # Contains all Nulls; this step not important though at all as this 'y' will have new notNull values \n",
    "del X_test_impute_marital_status['marital_status']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(52975, 22)"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_marital_status.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25394, 22)"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_marital_status.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_marital_status = logisticRegression_marital_status.predict(X_test_impute_marital_status)\n",
    "X_test_impute_marital_status['marital_status'] = predicted_marital_status #imputed @ null positions for marital_status"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'age_range' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "\n",
    "# Completing the X_train set :\n",
    "\n",
    "X_train_impute_marital_status['marital_status'] = y_train_marital_status\n",
    "\n",
    "X_train_impute_marital_status['rented'] = X_train_rented\n",
    "X_train_impute_marital_status['family_size'] = X_train_family_size\n",
    "X_train_impute_marital_status['no_of_children'] = X_train_no_of_children\n",
    "X_train_impute_marital_status['income_bracket'] = X_train_income_bracket\n",
    "X_train_impute_marital_status['redemption_status'] = X_train_redemption_status\n",
    "\n",
    "\n",
    "# Completing the X_test set :\n",
    "\n",
    "X_test_impute_marital_status['rented'] = X_test_rented\n",
    "X_test_impute_marital_status['family_size'] = X_test_family_size\n",
    "X_test_impute_marital_status['no_of_children'] = X_test_no_of_children\n",
    "X_test_impute_marital_status['income_bracket'] = X_test_income_bracket\n",
    "X_test_impute_marital_status['redemption_status'] = X_test_redemption_status"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (25394, 28)\n",
      "X_test shape : (52975, 28)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_marital_status.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_marital_status.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train_df shape : (78369, 28)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "train_df = X_train_impute_marital_status.append(X_test_impute_marital_status)\n",
    "print('Train_df shape :', train_df.shape)\n",
    "\n",
    "# ****'marital_status' feature IMPUTED ****"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='red'> 'rented' </font> for the train set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Encoding the 'marital_status' feature as 1/0 as it has been imputed successfully above :\n",
    "\n",
    "train_df['marital_status'] = train_df['marital_status'].apply(lambda x: 1 if x=='Married' else 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'age_range' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_rented = train_df.ix[train_df['rented'].notnull()] # where age_range is notNull\n",
    "\n",
    "# The following features are not considered for as a training set feature for 'age_range' imputation becoz they themselves contain Nulls \n",
    "\n",
    "X_train_redemption_status = X_train_impute_rented['redemption_status']\n",
    "del X_train_impute_rented['redemption_status']\n",
    "\n",
    "\n",
    "X_train_family_size = X_train_impute_rented['family_size']\n",
    "del X_train_impute_rented['family_size']\n",
    "\n",
    "X_train_no_of_children = X_train_impute_rented['no_of_children']\n",
    "del X_train_impute_rented['no_of_children']\n",
    "\n",
    "X_train_income_bracket = X_train_impute_rented['income_bracket'] \n",
    "del X_train_impute_rented['income_bracket']\n",
    "\n",
    "\n",
    "y_train_rented = X_train_impute_rented['rented']  # y variable i.e. age_range(with all notNull values)\n",
    "del X_train_impute_rented['rented']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(43661, 23)"
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_train_impute_age_range shape :\n",
    "\n",
    "X_train_impute_rented.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "                   intercept_scaling=1, l1_ratio=None, max_iter=100,\n",
       "                   multi_class='warn', n_jobs=None, penalty='l2',\n",
       "                   random_state=None, solver='warn', tol=0.0001, verbose=0,\n",
       "                   warm_start=False)"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Imputation using Model for 'age_range' features :\n",
    "\n",
    "logisticRegression_rented = LogisticRegression()\n",
    "logisticRegression_rented.fit(X_train_impute_rented, y_train_rented)  # LR fitted with X, y "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_rented contains\n",
    "\n",
    "X_test_impute_rented = train_df.ix[train_df['rented'].isnull()] # all the points containing 'marital_status' Nulls\n",
    "\n",
    "\n",
    "X_test_redemption_status = X_test_impute_rented['redemption_status']\n",
    "del X_test_impute_rented['redemption_status']\n",
    "\n",
    "X_test_family_size = X_test_impute_rented['family_size']\n",
    "del X_test_impute_rented['family_size']\n",
    "\n",
    "X_test_no_of_children = X_test_impute_rented['no_of_children']\n",
    "del X_test_impute_rented['no_of_children']\n",
    "\n",
    "X_test_income_bracket = X_test_impute_rented['income_bracket'] \n",
    "del X_test_impute_rented['income_bracket']\n",
    "\n",
    "\n",
    "y_test_rented = X_test_impute_rented['rented']  # Contains all Nulls; this step not important though at all as this 'y' will have new notNull values \n",
    "del X_test_impute_rented['rented']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(34708, 23)"
      ]
     },
     "execution_count": 99,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_rented.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(43661, 23)"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_rented.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_rented = logisticRegression_rented.predict(X_test_impute_rented)\n",
    "X_test_impute_rented['rented'] = predicted_rented"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'age_range' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "\n",
    "# Completing the X_train set :\n",
    "\n",
    "X_train_impute_rented['rented'] = y_train_rented\n",
    "\n",
    "X_train_impute_rented['family_size'] = X_train_family_size\n",
    "X_train_impute_rented['no_of_children'] = X_train_no_of_children\n",
    "X_train_impute_rented['income_bracket'] = X_train_income_bracket\n",
    "X_train_impute_rented['redemption_status'] = X_train_redemption_status\n",
    "\n",
    "\n",
    "# Completing the X_test set :\n",
    "\n",
    "X_test_impute_rented['family_size'] = X_test_family_size\n",
    "X_test_impute_rented['no_of_children'] = X_test_no_of_children\n",
    "X_test_impute_rented['income_bracket'] = X_test_income_bracket\n",
    "X_test_impute_rented['redemption_status'] = X_test_redemption_status"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (43661, 28)\n",
      "X_test shape : (34708, 28)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_rented.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_rented.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train_df shape : (78369, 28)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "train_df = X_train_impute_rented.append(X_test_impute_rented)\n",
    "print('Train_df shape :', train_df.shape)\n",
    "\n",
    "# **** 'rented' feature IMPUTED ****"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='red'> 'family_size' </font> for the train set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'family_size' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_family_size = train_df.ix[train_df['family_size'].notnull()] # where family_size is notNull\n",
    "\n",
    "\n",
    "X_train_redemption_status = X_train_impute_family_size['redemption_status']\n",
    "del X_train_impute_family_size['redemption_status']\n",
    "\n",
    "\n",
    "# The following features are not considered for as a training set feature for 'age_range' imputation becoz they themselves contain Nulls \n",
    "X_train_no_of_children = X_train_impute_family_size['no_of_children']\n",
    "del X_train_impute_family_size['no_of_children']\n",
    "\n",
    "X_train_income_bracket = X_train_impute_family_size['income_bracket'] \n",
    "del X_train_impute_family_size['income_bracket']\n",
    "\n",
    "\n",
    "y_train_family_size = X_train_impute_family_size['family_size']  # y variable i.e. family_size(with all notNull values)\n",
    "del X_train_impute_family_size['family_size']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(43661, 24)"
      ]
     },
     "execution_count": 106,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_train_impute_age_range shape :\n",
    "\n",
    "X_train_impute_family_size.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "                   intercept_scaling=1, l1_ratio=None, max_iter=100,\n",
       "                   multi_class='warn', n_jobs=None, penalty='l2',\n",
       "                   random_state=None, solver='warn', tol=0.0001, verbose=0,\n",
       "                   warm_start=False)"
      ]
     },
     "execution_count": 107,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Imputation using Model for 'family_size' features :\n",
    "\n",
    "logisticRegression_family_size = LogisticRegression()\n",
    "logisticRegression_family_size.fit(X_train_impute_family_size, y_train_family_size)  # LR fitted with X, y "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_family_size contains\n",
    "\n",
    "X_test_impute_family_size = train_df.ix[train_df['family_size'].isnull()] # all the points containing 'marial_status' Nulls\n",
    "\n",
    "\n",
    "X_test_redemption_status = X_test_impute_family_size['redemption_status']\n",
    "del X_test_impute_family_size['redemption_status']\n",
    "\n",
    "\n",
    "X_test_no_of_children = X_test_impute_family_size['no_of_children']\n",
    "del X_test_impute_family_size['no_of_children']\n",
    "\n",
    "X_test_income_bracket = X_test_impute_family_size['income_bracket'] \n",
    "del X_test_impute_family_size['income_bracket']\n",
    "\n",
    "\n",
    "y_test_family_size = X_test_impute_family_size['family_size']  # Contains all Nulls; this step not important though at all as this 'y' will have new notNull values \n",
    "del X_test_impute_family_size['family_size']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(34708, 24)"
      ]
     },
     "execution_count": 109,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_family_size.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(43661, 24)"
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_family_size.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_family_size = logisticRegression_family_size.predict(X_test_impute_family_size)\n",
    "X_test_impute_family_size['family_size'] = predicted_family_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'family_size' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "\n",
    "# Completing the X_train set :\n",
    "\n",
    "X_train_impute_family_size['family_size'] = y_train_family_size\n",
    "\n",
    "X_train_impute_family_size['no_of_children'] = X_train_no_of_children\n",
    "X_train_impute_family_size['income_bracket'] = X_train_income_bracket\n",
    "X_train_impute_family_size['redemption_status'] = X_train_redemption_status\n",
    "\n",
    "# Completing the X_test set :\n",
    "\n",
    "X_test_impute_family_size['no_of_children'] = X_test_no_of_children\n",
    "X_test_impute_family_size['income_bracket'] = X_test_income_bracket\n",
    "X_test_impute_family_size['redemption_status'] = X_test_redemption_status"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (43661, 28)\n",
      "X_test shape : (34708, 28)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_family_size.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_family_size.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train_df shape : (78369, 28)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "train_df = X_train_impute_family_size.append(X_test_impute_family_size)\n",
    "print('Train_df shape :', train_df.shape)\n",
    "\n",
    "# **** 'family_size' feature IMPUTED ****"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='red'> 'no_of_children' </font> for the train set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(78369, 32)"
      ]
     },
     "execution_count": 115,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# One hot encoding 'family_size' feature as it has been imputed successfully  :\n",
    "\n",
    "train_df = pd.concat([train_df, pd.get_dummies(train_df['family_size'])], axis=1)\n",
    "del train_df['family_size'] # has been one hot encoded, hence not required as a single feature\n",
    "train_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [],
   "source": [
    "# renaming family_size columns :\n",
    "\n",
    "train_df.rename(columns={'1':'family_size 1',\n",
    "                         '2':'family_size 2',\n",
    "                         '3':'family_size 3',\n",
    "                         '4':'family_size 4',\n",
    "                         '5+':'family_size 5+'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'no_of_children' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_no_of_children = train_df.ix[train_df['no_of_children'].notnull()] # where no_of_children is notNull\n",
    "\n",
    "\n",
    "X_train_redemption_status = X_train_impute_no_of_children['redemption_status'] \n",
    "del X_train_impute_no_of_children['redemption_status']\n",
    "\n",
    "# The following features are not considered for as a training set feature for 'no_of_children' imputation becoz they themselves contain Nulls \n",
    "\n",
    "X_train_income_bracket = X_train_impute_no_of_children['income_bracket'] \n",
    "del X_train_impute_no_of_children['income_bracket']\n",
    "\n",
    "\n",
    "y_train_no_of_children = X_train_impute_no_of_children['no_of_children']  # y variable i.e. no_of_children(with all notNull values)\n",
    "del X_train_impute_no_of_children['no_of_children']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(14056, 29)"
      ]
     },
     "execution_count": 118,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_train_impute_age_range shape :\n",
    "\n",
    "X_train_impute_no_of_children.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "                   intercept_scaling=1, l1_ratio=None, max_iter=100,\n",
       "                   multi_class='warn', n_jobs=None, penalty='l2',\n",
       "                   random_state=None, solver='warn', tol=0.0001, verbose=0,\n",
       "                   warm_start=False)"
      ]
     },
     "execution_count": 119,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Imputation using Model for 'family_size' features :\n",
    "\n",
    "logisticRegression_no_of_children = LogisticRegression()\n",
    "logisticRegression_no_of_children.fit(X_train_impute_no_of_children, y_train_no_of_children)  # LR fitted with X, y "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_no_of_children contains\n",
    "\n",
    "X_test_impute_no_of_children = train_df.ix[train_df['no_of_children'].isnull()] # all the points containing 'marial_status' Nulls\n",
    "\n",
    "\n",
    "X_test_redemption_status = X_test_impute_no_of_children['redemption_status'] \n",
    "del X_test_impute_no_of_children['redemption_status']\n",
    "\n",
    "\n",
    "X_test_income_bracket = X_test_impute_no_of_children['income_bracket'] \n",
    "del X_test_impute_no_of_children['income_bracket']\n",
    "\n",
    "\n",
    "y_test_no_of_children = X_test_impute_no_of_children['no_of_children']  # Contains all Nulls; this step not important though at all as this 'y' will have new notNull values \n",
    "del X_test_impute_no_of_children['no_of_children']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(64313, 29)"
      ]
     },
     "execution_count": 121,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_no_of_children.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(14056, 29)"
      ]
     },
     "execution_count": 122,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_no_of_children.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_no_of_children = logisticRegression_no_of_children.predict(X_test_impute_no_of_children)\n",
    "X_test_impute_no_of_children['no_of_children'] = predicted_no_of_children"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'family_size' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "\n",
    "# Completing the X_train set :\n",
    "\n",
    "X_train_impute_no_of_children['no_of_children'] = y_train_no_of_children\n",
    "\n",
    "X_train_impute_no_of_children['redemption_status'] = X_train_redemption_status\n",
    "X_train_impute_no_of_children['income_bracket'] = X_train_income_bracket\n",
    "\n",
    "\n",
    "# Completing the X_test set :\n",
    "\n",
    "X_test_impute_no_of_children['income_bracket'] = X_test_income_bracket\n",
    "X_test_impute_no_of_children['redemption_status'] = X_test_redemption_status"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (14056, 32)\n",
      "X_test shape : (64313, 32)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_no_of_children.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_no_of_children.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train_df shape : (78369, 32)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "train_df = X_train_impute_no_of_children.append(X_test_impute_no_of_children)\n",
    "print('Train_df shape :', train_df.shape)\n",
    "\n",
    "# **** 'no_of_children' feature IMPUTED ****"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='red'> 'income_bracket' </font> for the train set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(78369, 34)"
      ]
     },
     "execution_count": 127,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# One hot encoding 'family_size' feature as it has been imputed successfully  :\n",
    "\n",
    "train_df = pd.concat([train_df, pd.get_dummies(train_df['no_of_children'])], axis=1)\n",
    "del train_df['no_of_children'] # has been one hot encoded, hence not required as a single feature\n",
    "train_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [],
   "source": [
    "# renaming family_size columns :\n",
    "\n",
    "train_df.rename(columns={'1':'no_of_children 1',\n",
    "                         '2':'no_of_children 2',\n",
    "                         '3+':'no_of_children 3'},inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'income_bracket' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_income_bracket = train_df.ix[train_df['income_bracket'].notnull()] # where 'income_bracket' is notNull\n",
    "\n",
    "X_train_redemption_status= X_train_impute_income_bracket['redemption_status']\n",
    "del X_train_impute_income_bracket['redemption_status']\n",
    "\n",
    "y_train_income_bracket = X_train_impute_income_bracket['income_bracket']  # y variable i.e. 'income_bracket'(with all notNull values)\n",
    "del X_train_impute_income_bracket['income_bracket']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(43661, 32)"
      ]
     },
     "execution_count": 130,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_train_impute_age_range shape :\n",
    "\n",
    "X_train_impute_income_bracket.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "                   intercept_scaling=1, l1_ratio=None, max_iter=100,\n",
       "                   multi_class='warn', n_jobs=None, penalty='l2',\n",
       "                   random_state=None, solver='warn', tol=0.0001, verbose=0,\n",
       "                   warm_start=False)"
      ]
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Imputation using Model for 'family_size' features :\n",
    "\n",
    "logisticRegression_income_bracket = LogisticRegression()\n",
    "logisticRegression_income_bracket.fit(X_train_impute_income_bracket, y_train_income_bracket)  # LR fitted with X, y "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_income_bracket contains\n",
    "\n",
    "X_test_impute_income_bracket = train_df.ix[train_df['income_bracket'].isnull()] # all the points containing 'income_bracket' Nulls\n",
    "\n",
    "X_test_redemption_status = X_test_impute_income_bracket['redemption_status']\n",
    "del X_test_impute_income_bracket['redemption_status']\n",
    "\n",
    "\n",
    "y_test_income_bracket = X_test_impute_income_bracket['income_bracket']  # Contains all Nulls; this step not important though at all as this 'y' will have new notNull values \n",
    "del X_test_impute_income_bracket['income_bracket']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(34708, 32)"
      ]
     },
     "execution_count": 133,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_income_bracket.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(43661, 32)"
      ]
     },
     "execution_count": 134,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_income_bracket.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_income_bracket = logisticRegression_income_bracket.predict(X_test_impute_income_bracket)\n",
    "X_test_impute_income_bracket['income_bracket'] = predicted_income_bracket"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 136,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'income_bracket' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "# Completing the X_train set :\n",
    "\n",
    "X_train_impute_income_bracket['income_bracket'] = y_train_income_bracket\n",
    "X_train_impute_income_bracket['redemption_status'] = X_train_redemption_status\n",
    "\n",
    "# Completing the X_test set :\n",
    "\n",
    "X_test_impute_income_bracket['redemption_status'] = X_test_redemption_status "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (43661, 34)\n",
      "X_test shape : (34708, 34)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_income_bracket.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_income_bracket.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train_df shape : (78369, 34)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "train_df = X_train_impute_income_bracket.append(X_test_impute_income_bracket)\n",
    "print('Train_df shape :', train_df.shape)\n",
    "\n",
    "# **** 'income_bracket' feature IMPUTED ****"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "18-25                   0\n",
       "26-35                   0\n",
       "36-45                   0\n",
       "46-55                   0\n",
       "56-70                   0\n",
       "70+                     0\n",
       "Gap_days_campaign       0\n",
       "X                       0\n",
       "Y                       0\n",
       "category                0\n",
       "coupon_discount_max     0\n",
       "coupon_discount_mean    0\n",
       "coupon_discount_min     0\n",
       "coupon_discount_sum     0\n",
       "family_size 1           0\n",
       "family_size 2           0\n",
       "family_size 3           0\n",
       "family_size 4           0\n",
       "family_size 5+          0\n",
       "marital_status          0\n",
       "other_discount_max      0\n",
       "other_discount_mean     0\n",
       "other_discount_min      0\n",
       "other_discount_sum      0\n",
       "rented                  0\n",
       "selling_price_max       0\n",
       "selling_price_mean      0\n",
       "selling_price_min       0\n",
       "selling_price_sum       0\n",
       "no_of_children 1        0\n",
       "no_of_children 2        0\n",
       "no_of_children 3        0\n",
       "income_bracket          0\n",
       "redemption_status       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 139,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# All the nulls have been imputed successfully and the models used for the imputation shall be utilized in the test set as well!!\n",
    "\n",
    "train_df.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h3> Test set features to impute :</h3>\n",
    "\n",
    "+ age_range               \n",
    "+ marital_status          \n",
    "+ rented                  \n",
    "+ family_size             \n",
    "+ no_of_children          \n",
    "+ income_bracket          "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 234,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----------------+-------------------+-----------+----------------+--------+-------------+----------------+----------------+\n",
      "| FeatureToImpute | redemption_status | age_range | marital_status | rented | family_size | no_of_children | income_Bracket |\n",
      "+-----------------+-------------------+-----------+----------------+--------+-------------+----------------+----------------+\n",
      "|    age_range    |        N/A        |    N/A    |       X        |   X    |      X      |       X        |       X        |\n",
      "|  marital_status |        N/A        |     ✓     |      N/A       |   X    |      X      |       X        |       X        |\n",
      "|      rented     |        N/A        |     ✓     |       ✓        |  N/A   |      X      |       X        |       X        |\n",
      "|   family_size   |        N/A        |     ✓     |       ✓        |   ✓    |     N/A     |       X        |       X        |\n",
      "|  no_of_children |        N/A        |     ✓     |       ✓        |   ✓    |      ✓      |      N/A       |       X        |\n",
      "|  income_bracket |        N/A        |     ✓     |       ✓        |   ✓    |      ✓      |       ✓        |      N/A       |\n",
      "+-----------------+-------------------+-----------+----------------+--------+-------------+----------------+----------------+\n"
     ]
    }
   ],
   "source": [
    "# Utilization of the features in incremental way to achieve imputation in the 'after' features : \n",
    "# The models will be used exactly in the same step as they were used for imputation in the train set :\n",
    "\n",
    "table = PrettyTable()\n",
    "table.field_names = ['FeatureToImpute', 'redemption_status','age_range', 'marital_status', 'rented', 'family_size', 'no_of_children', 'income_Bracket']\n",
    "\n",
    "table.align[\"FeatureToImpute \"] = \"L\"\n",
    "table.align[\"redemption_status\"] = \"L\"\n",
    "table.align[\"age_range\"] = \"R\"\n",
    "table.align[\"marital_status \"] = \"R\"\n",
    "table.align[\"rented\"] = \"R\"\n",
    "table.align[\"family_size\"] = \"R\"\n",
    "table.align[\"no_of_children\"] = \"R\"\n",
    "table.align[\"income_bracket\"] = \"L\"\n",
    "\n",
    "\n",
    "table.add_row(['age_range','N/A', 'N/A', 'X' , 'X' , 'X' , 'X' , 'X'  ])\n",
    "\n",
    "table.add_row(['marital_status','N/A', '✓', 'N/A', 'X' , 'X' , 'X' , 'X' ])\n",
    "\n",
    "table.add_row(['rented','N/A', '✓', '✓', 'N/A', 'X' , 'X' , 'X' ])\n",
    "\n",
    "table.add_row(['family_size','N/A', '✓', '✓', '✓', 'N/A', 'X' , 'X' ])\n",
    "\n",
    "table.add_row(['no_of_children', 'N/A', '✓', '✓', '✓', '✓', 'N/A', 'X' ])\n",
    "\n",
    "table.add_row(['income_bracket', 'N/A', '✓', '✓', '✓', '✓', '✓', 'N/A'])\n",
    "\n",
    "print(table)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='blue'> 'age_range' </font> for the test set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50226, 22)"
      ]
     },
     "execution_count": 141,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# One hot encoding 'campaign_type' feature :\n",
    "\n",
    "test_df = pd.concat([test_df, pd.get_dummies(test_df['campaign_type'])], axis=1)\n",
    "del test_df['campaign_type'] # has been one hot encoded, hence not required as a single feature\n",
    "test_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'age_range' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_age_range = test_df.ix[test_df['age_range'].notnull()] # where age_range is notNull\n",
    "\n",
    "# The following features are not considered for as a training set feature for 'age_range' imputation becoz they themselves contain Nulls \n",
    "\n",
    "X_train_marital_status = X_train_impute_age_range['marital_status']\n",
    "del X_train_impute_age_range['marital_status']\n",
    "\n",
    "X_train_rented = X_train_impute_age_range['rented']\n",
    "del X_train_impute_age_range['rented']\n",
    "\n",
    "X_train_family_size = X_train_impute_age_range['family_size']\n",
    "del X_train_impute_age_range['family_size']\n",
    "\n",
    "X_train_no_of_children = X_train_impute_age_range['no_of_children']\n",
    "del X_train_impute_age_range['no_of_children']\n",
    "\n",
    "X_train_income_bracket = X_train_impute_age_range['income_bracket'] \n",
    "del X_train_impute_age_range['income_bracket']\n",
    "\n",
    "\n",
    "y_train_age_range = X_train_impute_age_range['age_range']  # y variable i.e. age_range(with all notNull values)\n",
    "del X_train_impute_age_range['age_range']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_age_range contains\n",
    "\n",
    "X_test_impute_age_range = test_df.ix[test_df['age_range'].isnull()] # all the points containing 'age_range' Nulls\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "X_test_marital_status = X_test_impute_age_range['marital_status']\n",
    "del X_test_impute_age_range['marital_status']\n",
    "\n",
    "X_test_rented = X_test_impute_age_range['rented']\n",
    "del X_test_impute_age_range['rented']\n",
    "\n",
    "X_test_family_size = X_test_impute_age_range['family_size']\n",
    "del X_test_impute_age_range['family_size']\n",
    "\n",
    "X_test_no_of_children = X_test_impute_age_range['no_of_children']\n",
    "del X_test_impute_age_range['no_of_children']\n",
    "\n",
    "X_test_income_bracket = X_test_impute_age_range['income_bracket'] \n",
    "del X_test_impute_age_range['income_bracket']\n",
    "\n",
    "# Contains all Nulls; this step not important though at all as this 'y' will have new not null values \n",
    "y_test_age_range = X_test_impute_age_range['age_range']  \n",
    "del X_test_impute_age_range['age_range']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "metadata": {},
   "outputs": [],
   "source": [
    "# direct prediction from the already trained model for the feature 'age_range':\n",
    "\n",
    "predicted_age_range = logisticRegression_age_range.predict(X_test_impute_age_range)\n",
    "X_test_impute_age_range['age_range'] = predicted_age_range # imputation via predicted values for the nulls"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'age_range' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "\n",
    "# Completing the X_train set :\n",
    "X_train_impute_age_range['age_range'] = y_train_age_range\n",
    "X_train_impute_age_range['marital_status'] = X_train_marital_status\n",
    "X_train_impute_age_range['rented'] = X_train_rented\n",
    "X_train_impute_age_range['family_size'] = X_train_family_size\n",
    "X_train_impute_age_range['no_of_children'] = X_train_no_of_children\n",
    "X_train_impute_age_range['income_bracket'] = X_train_income_bracket\n",
    "\n",
    "\n",
    "# Completing the X_test set :\n",
    "# X_test_impute_age_range['age_range'] = predicted_age_range # imputation via predicted values for the nulls # already imputed in above cell\n",
    "X_test_impute_age_range['marital_status'] = X_test_marital_status \n",
    "X_test_impute_age_range['rented'] = X_test_rented\n",
    "X_test_impute_age_range['family_size'] = X_test_family_size\n",
    "X_test_impute_age_range['no_of_children'] = X_test_no_of_children\n",
    "X_test_impute_age_range['income_bracket'] = X_test_income_bracket"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (30939, 22)\n",
      "X_test shape : (19287, 22)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_age_range.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_age_range.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_df shape : (50226, 22)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "test_df = X_train_impute_age_range.append(X_test_impute_age_range)\n",
    "print('test_df shape :', test_df.shape)\n",
    "\n",
    "# 'age_range' feature IMPUTED *******"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='blue'> 'marital_status' </font> for the test set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50226, 27)"
      ]
     },
     "execution_count": 148,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# One hot encoding 'campaign_type' feature :\n",
    "\n",
    "test_df = pd.concat([test_df, pd.get_dummies(test_df['age_range'])], axis=1)\n",
    "del test_df['age_range'] # has been one hot encoded, hence not required as a single feature\n",
    "test_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'age_range' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_marital_status = test_df.ix[test_df['marital_status'].notnull()] # where age_range is notNull\n",
    "\n",
    "\n",
    "\n",
    "# The following features are not considered for as a training set feature for 'age_range' imputation becoz they themselves contain Nulls \n",
    "\n",
    "X_train_rented = X_train_impute_marital_status['rented']\n",
    "del X_train_impute_marital_status['rented']\n",
    "\n",
    "X_train_family_size = X_train_impute_marital_status['family_size']\n",
    "del X_train_impute_marital_status['family_size']\n",
    "\n",
    "X_train_no_of_children = X_train_impute_marital_status['no_of_children']\n",
    "del X_train_impute_marital_status['no_of_children']\n",
    "\n",
    "X_train_income_bracket = X_train_impute_marital_status['income_bracket'] \n",
    "del X_train_impute_marital_status['income_bracket']\n",
    "\n",
    "\n",
    "y_train_marital_status = X_train_impute_marital_status['marital_status']  # y variable i.e. age_range(with all notNull values)\n",
    "del X_train_impute_marital_status['marital_status']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(18082, 22)"
      ]
     },
     "execution_count": 150,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_train_impute_age_range shape :\n",
    "\n",
    "X_train_impute_marital_status.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_marital_status contains\n",
    "\n",
    "X_test_impute_marital_status = test_df.ix[test_df['marital_status'].isnull()] # all the points containing 'marial_status' Nulls\n",
    "\n",
    "\n",
    "X_test_rented = X_test_impute_marital_status['rented']\n",
    "del X_test_impute_marital_status['rented']\n",
    "\n",
    "X_test_family_size = X_test_impute_marital_status['family_size']\n",
    "del X_test_impute_marital_status['family_size']\n",
    "\n",
    "X_test_no_of_children = X_test_impute_marital_status['no_of_children']\n",
    "del X_test_impute_marital_status['no_of_children']\n",
    "\n",
    "X_test_income_bracket = X_test_impute_marital_status['income_bracket'] \n",
    "del X_test_impute_marital_status['income_bracket']\n",
    "\n",
    "\n",
    "y_test_marital_status = X_test_impute_marital_status['marital_status']  # Contains all Nulls; this step not important though at all as this 'y' will have new notNull values \n",
    "del X_test_impute_marital_status['marital_status']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(32144, 22)"
      ]
     },
     "execution_count": 152,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_marital_status.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(18082, 22)"
      ]
     },
     "execution_count": 153,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_marital_status.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_marital_status = logisticRegression_marital_status.predict(X_test_impute_marital_status)\n",
    "X_test_impute_marital_status['marital_status'] = predicted_marital_status #imputed @ null positions for marital_status"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'age_range' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "\n",
    "# Completing the X_train set :\n",
    "\n",
    "X_train_impute_marital_status['marital_status'] = y_train_marital_status\n",
    "\n",
    "X_train_impute_marital_status['rented'] = X_train_rented\n",
    "X_train_impute_marital_status['family_size'] = X_train_family_size\n",
    "X_train_impute_marital_status['no_of_children'] = X_train_no_of_children\n",
    "X_train_impute_marital_status['income_bracket'] = X_train_income_bracket\n",
    "\n",
    "\n",
    "# Completing the X_test set :\n",
    "\n",
    "X_test_impute_marital_status['rented'] = X_test_rented\n",
    "X_test_impute_marital_status['family_size'] = X_test_family_size\n",
    "X_test_impute_marital_status['no_of_children'] = X_test_no_of_children\n",
    "X_test_impute_marital_status['income_bracket'] = X_test_income_bracket\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (18082, 27)\n",
      "X_test shape : (32144, 27)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_marital_status.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_marital_status.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_df shape : (50226, 27)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "test_df = X_train_impute_marital_status.append(X_test_impute_marital_status)\n",
    "print('test_df shape :', test_df.shape)\n",
    "\n",
    "# ****'marital_status' feature IMPUTED ****"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='blue'> 'rented' </font> for the test set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Encoding the 'marital_status' feature as 1/0 as it has been imputed successfully above :\n",
    "\n",
    "test_df['marital_status'] = test_df['marital_status'].apply(lambda x: 1 if x=='Married' else 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'age_range' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_rented = test_df.ix[test_df['rented'].notnull()] # where age_range is notNull\n",
    "\n",
    "# The following features are not considered for as a training set feature for 'age_range' imputation becoz they themselves contain Nulls \n",
    "\n",
    "\n",
    "X_train_family_size = X_train_impute_rented['family_size']\n",
    "del X_train_impute_rented['family_size']\n",
    "\n",
    "X_train_no_of_children = X_train_impute_rented['no_of_children']\n",
    "del X_train_impute_rented['no_of_children']\n",
    "\n",
    "X_train_income_bracket = X_train_impute_rented['income_bracket'] \n",
    "del X_train_impute_rented['income_bracket']\n",
    "\n",
    "\n",
    "y_train_rented = X_train_impute_rented['rented']  # y variable i.e. age_range(with all notNull values)\n",
    "del X_train_impute_rented['rented']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30939, 23)"
      ]
     },
     "execution_count": 160,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_train_impute_age_range shape :\n",
    "\n",
    "X_train_impute_rented.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_rented contains\n",
    "\n",
    "X_test_impute_rented = test_df.ix[test_df['rented'].isnull()] # all the points containing 'marital_status' Nulls\n",
    "\n",
    "\n",
    "X_test_family_size = X_test_impute_rented['family_size']\n",
    "del X_test_impute_rented['family_size']\n",
    "\n",
    "X_test_no_of_children = X_test_impute_rented['no_of_children']\n",
    "del X_test_impute_rented['no_of_children']\n",
    "\n",
    "X_test_income_bracket = X_test_impute_rented['income_bracket'] \n",
    "del X_test_impute_rented['income_bracket']\n",
    "\n",
    "\n",
    "y_test_rented = X_test_impute_rented['rented']  # Contains all Nulls; this step not important though at all as this 'y' will have new notNull values \n",
    "del X_test_impute_rented['rented']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(19287, 23)"
      ]
     },
     "execution_count": 162,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_rented.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30939, 23)"
      ]
     },
     "execution_count": 163,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_rented.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_rented = logisticRegression_rented.predict(X_test_impute_rented)\n",
    "X_test_impute_rented['rented'] = predicted_rented"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'age_range' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "\n",
    "# Completing the X_train set :\n",
    "\n",
    "X_train_impute_rented['rented'] = y_train_rented\n",
    "\n",
    "X_train_impute_rented['family_size'] = X_train_family_size\n",
    "X_train_impute_rented['no_of_children'] = X_train_no_of_children\n",
    "X_train_impute_rented['income_bracket'] = X_train_income_bracket\n",
    "\n",
    "\n",
    "# Completing the X_test set :\n",
    "\n",
    "X_test_impute_rented['family_size'] = X_test_family_size\n",
    "X_test_impute_rented['no_of_children'] = X_test_no_of_children\n",
    "X_test_impute_rented['income_bracket'] = X_test_income_bracket"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (30939, 27)\n",
      "X_test shape : (19287, 27)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_rented.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_rented.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_df shape : (50226, 27)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "test_df = X_train_impute_rented.append(X_test_impute_rented)\n",
    "print('test_df shape :', test_df.shape)\n",
    "\n",
    "# **** 'rented' feature IMPUTED ****"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='blue'> 'family_size' </font> for the test set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'family_size' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_family_size = test_df.ix[test_df['family_size'].notnull()] # where family_size is notNull\n",
    "\n",
    "\n",
    "# The following features are not considered for as a training set feature for 'age_range' imputation becoz they themselves contain Nulls \n",
    "X_train_no_of_children = X_train_impute_family_size['no_of_children']\n",
    "del X_train_impute_family_size['no_of_children']\n",
    "\n",
    "X_train_income_bracket = X_train_impute_family_size['income_bracket'] \n",
    "del X_train_impute_family_size['income_bracket']\n",
    "\n",
    "\n",
    "y_train_family_size = X_train_impute_family_size['family_size']  # y variable i.e. family_size(with all notNull values)\n",
    "del X_train_impute_family_size['family_size']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30939, 24)"
      ]
     },
     "execution_count": 169,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_train_impute_family_size shape :\n",
    "\n",
    "X_train_impute_family_size.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_family_size contains\n",
    "\n",
    "X_test_impute_family_size = test_df.ix[test_df['family_size'].isnull()] # all the points containing 'marial_status' Nulls\n",
    "\n",
    "\n",
    "X_test_no_of_children = X_test_impute_family_size['no_of_children']\n",
    "del X_test_impute_family_size['no_of_children']\n",
    "\n",
    "X_test_income_bracket = X_test_impute_family_size['income_bracket'] \n",
    "del X_test_impute_family_size['income_bracket']\n",
    "\n",
    "\n",
    "y_test_family_size = X_test_impute_family_size['family_size']  # Contains all Nulls; this step not important though at all as this 'y' will have new notNull values \n",
    "del X_test_impute_family_size['family_size']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(19287, 24)"
      ]
     },
     "execution_count": 171,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_family_size.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30939, 24)"
      ]
     },
     "execution_count": 172,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_family_size.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_family_size = logisticRegression_family_size.predict(X_test_impute_family_size)\n",
    "X_test_impute_family_size['family_size'] = predicted_family_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'family_size' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "\n",
    "# Completing the X_train set :\n",
    "\n",
    "X_train_impute_family_size['family_size'] = y_train_family_size\n",
    "\n",
    "X_train_impute_family_size['no_of_children'] = X_train_no_of_children\n",
    "X_train_impute_family_size['income_bracket'] = X_train_income_bracket\n",
    "\n",
    "# Completing the X_test set :\n",
    "\n",
    "X_test_impute_family_size['no_of_children'] = X_test_no_of_children\n",
    "X_test_impute_family_size['income_bracket'] = X_test_income_bracket"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (30939, 27)\n",
      "X_test shape : (19287, 27)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_family_size.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_family_size.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_df shape : (50226, 27)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "test_df = X_train_impute_family_size.append(X_test_impute_family_size)\n",
    "print('test_df shape :', test_df.shape)\n",
    "\n",
    "# **** 'family_size' feature IMPUTED ****"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='blue'> 'no_of_children' </font> for the test set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50226, 31)"
      ]
     },
     "execution_count": 177,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# One hot encoding 'family_size' feature as it has been imputed successfully  :\n",
    "\n",
    "test_df = pd.concat([test_df, pd.get_dummies(test_df['family_size'])], axis=1)\n",
    "del test_df['family_size'] # has been one hot encoded, hence not required as a single feature\n",
    "test_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {},
   "outputs": [],
   "source": [
    "# renaming family_size columns :\n",
    "\n",
    "test_df.rename(columns={'1':'family_size 1',\n",
    "                         '2':'family_size 2',\n",
    "                         '3':'family_size 3',\n",
    "                         '4':'family_size 4',\n",
    "                         '5+':'family_size 5+'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'no_of_children' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_no_of_children = test_df.ix[test_df['no_of_children'].notnull()] # where no_of_children is notNull\n",
    "\n",
    "# The following features are not considered for as a training set feature for 'no_of_children' imputation becoz they themselves contain Nulls \n",
    "\n",
    "X_train_income_bracket = X_train_impute_no_of_children['income_bracket'] \n",
    "del X_train_impute_no_of_children['income_bracket']\n",
    "\n",
    "\n",
    "y_train_no_of_children = X_train_impute_no_of_children['no_of_children']  # y variable i.e. no_of_children(with all notNull values)\n",
    "del X_train_impute_no_of_children['no_of_children']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10162, 29)"
      ]
     },
     "execution_count": 180,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_train_impute_age_range shape :\n",
    "\n",
    "X_train_impute_no_of_children.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_no_of_children contains\n",
    "\n",
    "X_test_impute_no_of_children = test_df.ix[test_df['no_of_children'].isnull()] # all the points containing 'marial_status' Nulls\n",
    "\n",
    "\n",
    "X_test_income_bracket = X_test_impute_no_of_children['income_bracket'] \n",
    "del X_test_impute_no_of_children['income_bracket']\n",
    "\n",
    "\n",
    "y_test_no_of_children = X_test_impute_no_of_children['no_of_children']  # Contains all Nulls; this step not important though at all as this 'y' will have new notNull values \n",
    "del X_test_impute_no_of_children['no_of_children']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(40064, 29)"
      ]
     },
     "execution_count": 182,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_no_of_children.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10162, 29)"
      ]
     },
     "execution_count": 183,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_no_of_children.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_no_of_children = logisticRegression_no_of_children.predict(X_test_impute_no_of_children)\n",
    "X_test_impute_no_of_children['no_of_children'] = predicted_no_of_children"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'family_size' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "\n",
    "# Completing the X_train set :\n",
    "\n",
    "X_train_impute_no_of_children['no_of_children'] = y_train_no_of_children\n",
    "\n",
    "X_train_impute_no_of_children['income_bracket'] = X_train_income_bracket\n",
    "\n",
    "\n",
    "# Completing the X_test set :\n",
    "\n",
    "X_test_impute_no_of_children['income_bracket'] = X_test_income_bracket"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (10162, 31)\n",
      "X_test shape : (40064, 31)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_no_of_children.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_no_of_children.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_df shape : (50226, 31)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "test_df = X_train_impute_no_of_children.append(X_test_impute_no_of_children)\n",
    "print('test_df shape :', test_df.shape)\n",
    "\n",
    "# **** 'no_of_children' feature IMPUTED ****"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> <h2>Imputing feature <font color='blue'> 'income_bracket' </font> for the test set :</h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50226, 33)"
      ]
     },
     "execution_count": 188,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# One hot encoding 'family_size' feature as it has been imputed successfully  :\n",
    "\n",
    "test_df = pd.concat([test_df, pd.get_dummies(test_df['no_of_children'])], axis=1)\n",
    "del test_df['no_of_children'] # has been one hot encoded, hence not required as a single feature\n",
    "test_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {},
   "outputs": [],
   "source": [
    "# renaming family_size columns :\n",
    "\n",
    "test_df.rename(columns={'1':'no_of_children 1',\n",
    "                         '2':'no_of_children 2',\n",
    "                         '3+':'no_of_children 3'},inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate dataframe containing each features null and feature 'income_bracket' removed from the above created dataset :\n",
    "\n",
    "X_train_impute_income_bracket = test_df.ix[test_df['income_bracket'].notnull()] # where 'income_bracket' is notNull\n",
    "\n",
    "y_train_income_bracket = X_train_impute_income_bracket['income_bracket']  # y variable i.e. 'income_bracket'(with all notNull values)\n",
    "del X_train_impute_income_bracket['income_bracket']  # removing the y variable from the learning_set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30939, 32)"
      ]
     },
     "execution_count": 191,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# X_train_impute_age_range shape :\n",
    "\n",
    "X_train_impute_income_bracket.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TO PREDICT ON : Where 'age_range' feature is null :\n",
    "# X_test_impute_income_bracket contains\n",
    "\n",
    "X_test_impute_income_bracket = test_df.ix[test_df['income_bracket'].isnull()] # all the points containing 'income_bracket' Nulls\n",
    "\n",
    "y_test_income_bracket = X_test_impute_income_bracket['income_bracket']  # Contains all Nulls; this step not important though at all as this 'y' will have new notNull values \n",
    "del X_test_impute_income_bracket['income_bracket']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(19287, 32)"
      ]
     },
     "execution_count": 193,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test_impute_income_bracket.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 194,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30939, 32)"
      ]
     },
     "execution_count": 194,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_impute_income_bracket.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_income_bracket = logisticRegression_income_bracket.predict(X_test_impute_income_bracket)\n",
    "X_test_impute_income_bracket['income_bracket'] = predicted_income_bracket"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combining X_train and X_test sets after 'income_bracket' feature imputation :\n",
    "# Also including the removed features in the above imputation procedure :\n",
    "\n",
    "# Completing the X_train set :\n",
    "\n",
    "X_train_impute_income_bracket['income_bracket'] = y_train_income_bracket\n",
    "\n",
    "# Completing the X_test set :\n",
    "# Already accomplished"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape : (30939, 33)\n",
      "X_test shape : (19287, 33)\n"
     ]
    }
   ],
   "source": [
    "print(\"X_train shape :\" , X_train_impute_income_bracket.shape)\n",
    "print(\"X_test shape :\" , X_test_impute_income_bracket.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_df shape : (50226, 33)\n"
     ]
    }
   ],
   "source": [
    "# Combining the X_train & X_test :\n",
    "\n",
    "test_df = X_train_impute_income_bracket.append(X_test_impute_income_bracket)\n",
    "print('test_df shape :', test_df.shape)\n",
    "\n",
    "# **** 'income_bracket' feature IMPUTED ****"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### *Final train & test sets dimensionality :*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(78369, 34)"
      ]
     },
     "execution_count": 199,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# One xtra (, 5) dimension beoz of feature 'redemption_status' available for train set by default :\n",
    "\n",
    "train_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'redemption_status'}"
      ]
     },
     "execution_count": 206,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 'train_df' is one feature more to 'test_df' which is 'redemption_status'\n",
    "\n",
    "set(train_df.columns) - set(test_df.columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df.to_csv(r\"D:\\MOOCs & Material\\Data Science & ML\\Data Modelling - Datasets\\American Express (AmExpert) - Analytics vidhya\\Train & Test imputed\\train.csv\", index=False)\n",
    "test_df.to_csv(r\"D:\\MOOCs & Material\\Data Science & ML\\Data Modelling - Datasets\\American Express (AmExpert) - Analytics vidhya\\Train & Test imputed\\test.csv\", index=False)"
   ]
  },
  {
   "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}