Task2_Exploratory Data Analysis and Data Cleaning.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h1>Exploratory Data Analysis & Data Cleaning </h1>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#import libraries\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Load the dataset\n",
    "\n",
    "customer_data = pd.read_csv('Data\\ml_case_training_data.csv')\n",
    "pricing_data = pd.read_csv('Data\\ml_case_training_hist_data.csv')\n",
    "churn_status = pd.read_csv('Data\\ml_case_training_output.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>activity_new</th>\n",
       "      <th>campaign_disc_ele</th>\n",
       "      <th>channel_sales</th>\n",
       "      <th>cons_12m</th>\n",
       "      <th>cons_gas_12m</th>\n",
       "      <th>cons_last_month</th>\n",
       "      <th>date_activ</th>\n",
       "      <th>date_end</th>\n",
       "      <th>date_first_activ</th>\n",
       "      <th>...</th>\n",
       "      <th>forecast_price_pow_p1</th>\n",
       "      <th>has_gas</th>\n",
       "      <th>imp_cons</th>\n",
       "      <th>margin_gross_pow_ele</th>\n",
       "      <th>margin_net_pow_ele</th>\n",
       "      <th>nb_prod_act</th>\n",
       "      <th>net_margin</th>\n",
       "      <th>num_years_antig</th>\n",
       "      <th>origin_up</th>\n",
       "      <th>pow_max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>48ada52261e7cf58715202705a0451c9</td>\n",
       "      <td>esoiiifxdlbkcsluxmfuacbdckommixw</td>\n",
       "      <td>NaN</td>\n",
       "      <td>lmkebamcaaclubfxadlmueccxoimlema</td>\n",
       "      <td>309275</td>\n",
       "      <td>0</td>\n",
       "      <td>10025</td>\n",
       "      <td>2012-11-07</td>\n",
       "      <td>2016-11-06</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>58.995952</td>\n",
       "      <td>f</td>\n",
       "      <td>831.8</td>\n",
       "      <td>-41.76</td>\n",
       "      <td>-41.76</td>\n",
       "      <td>1</td>\n",
       "      <td>1732.36</td>\n",
       "      <td>3</td>\n",
       "      <td>ldkssxwpmemidmecebumciepifcamkci</td>\n",
       "      <td>180.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>24011ae4ebbe3035111d65fa7c15bc57</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>foosdfpfkusacimwkcsosbicdxkicaua</td>\n",
       "      <td>0</td>\n",
       "      <td>54946</td>\n",
       "      <td>0</td>\n",
       "      <td>2013-06-15</td>\n",
       "      <td>2016-06-15</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>40.606701</td>\n",
       "      <td>t</td>\n",
       "      <td>0.0</td>\n",
       "      <td>25.44</td>\n",
       "      <td>25.44</td>\n",
       "      <td>2</td>\n",
       "      <td>678.99</td>\n",
       "      <td>3</td>\n",
       "      <td>lxidpiddsbxsbosboudacockeimpuepw</td>\n",
       "      <td>43.648</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>d29c2c54acc38ff3c0614d0a653813dd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4660</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2009-08-21</td>\n",
       "      <td>2016-08-30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>44.311378</td>\n",
       "      <td>f</td>\n",
       "      <td>0.0</td>\n",
       "      <td>16.38</td>\n",
       "      <td>16.38</td>\n",
       "      <td>1</td>\n",
       "      <td>18.89</td>\n",
       "      <td>6</td>\n",
       "      <td>kamkkxfxxuwbdslkwifmmcsiusiuosws</td>\n",
       "      <td>13.800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>764c75f661154dac3a6c254cd082ea7d</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>foosdfpfkusacimwkcsosbicdxkicaua</td>\n",
       "      <td>544</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2010-04-16</td>\n",
       "      <td>2016-04-16</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>44.311378</td>\n",
       "      <td>f</td>\n",
       "      <td>0.0</td>\n",
       "      <td>28.60</td>\n",
       "      <td>28.60</td>\n",
       "      <td>1</td>\n",
       "      <td>6.60</td>\n",
       "      <td>6</td>\n",
       "      <td>kamkkxfxxuwbdslkwifmmcsiusiuosws</td>\n",
       "      <td>13.856</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>bba03439a292a1e166f80264c16191cb</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>lmkebamcaaclubfxadlmueccxoimlema</td>\n",
       "      <td>1584</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2010-03-30</td>\n",
       "      <td>2016-03-30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>44.311378</td>\n",
       "      <td>f</td>\n",
       "      <td>0.0</td>\n",
       "      <td>30.22</td>\n",
       "      <td>30.22</td>\n",
       "      <td>1</td>\n",
       "      <td>25.46</td>\n",
       "      <td>6</td>\n",
       "      <td>kamkkxfxxuwbdslkwifmmcsiusiuosws</td>\n",
       "      <td>13.200</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 32 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 id                      activity_new  \\\n",
       "0  48ada52261e7cf58715202705a0451c9  esoiiifxdlbkcsluxmfuacbdckommixw   \n",
       "1  24011ae4ebbe3035111d65fa7c15bc57                               NaN   \n",
       "2  d29c2c54acc38ff3c0614d0a653813dd                               NaN   \n",
       "3  764c75f661154dac3a6c254cd082ea7d                               NaN   \n",
       "4  bba03439a292a1e166f80264c16191cb                               NaN   \n",
       "\n",
       "   campaign_disc_ele                     channel_sales  cons_12m  \\\n",
       "0                NaN  lmkebamcaaclubfxadlmueccxoimlema    309275   \n",
       "1                NaN  foosdfpfkusacimwkcsosbicdxkicaua         0   \n",
       "2                NaN                               NaN      4660   \n",
       "3                NaN  foosdfpfkusacimwkcsosbicdxkicaua       544   \n",
       "4                NaN  lmkebamcaaclubfxadlmueccxoimlema      1584   \n",
       "\n",
       "   cons_gas_12m  cons_last_month  date_activ    date_end date_first_activ  \\\n",
       "0             0            10025  2012-11-07  2016-11-06              NaN   \n",
       "1         54946                0  2013-06-15  2016-06-15              NaN   \n",
       "2             0                0  2009-08-21  2016-08-30              NaN   \n",
       "3             0                0  2010-04-16  2016-04-16              NaN   \n",
       "4             0                0  2010-03-30  2016-03-30              NaN   \n",
       "\n",
       "   ... forecast_price_pow_p1 has_gas  imp_cons  margin_gross_pow_ele  \\\n",
       "0  ...             58.995952       f     831.8                -41.76   \n",
       "1  ...             40.606701       t       0.0                 25.44   \n",
       "2  ...             44.311378       f       0.0                 16.38   \n",
       "3  ...             44.311378       f       0.0                 28.60   \n",
       "4  ...             44.311378       f       0.0                 30.22   \n",
       "\n",
       "   margin_net_pow_ele  nb_prod_act  net_margin  num_years_antig  \\\n",
       "0              -41.76            1     1732.36                3   \n",
       "1               25.44            2      678.99                3   \n",
       "2               16.38            1       18.89                6   \n",
       "3               28.60            1        6.60                6   \n",
       "4               30.22            1       25.46                6   \n",
       "\n",
       "                          origin_up  pow_max  \n",
       "0  ldkssxwpmemidmecebumciepifcamkci  180.000  \n",
       "1  lxidpiddsbxsbosboudacockeimpuepw   43.648  \n",
       "2  kamkkxfxxuwbdslkwifmmcsiusiuosws   13.800  \n",
       "3  kamkkxfxxuwbdslkwifmmcsiusiuosws   13.856  \n",
       "4  kamkkxfxxuwbdslkwifmmcsiusiuosws   13.200  \n",
       "\n",
       "[5 rows x 32 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 16096 entries, 0 to 16095\n",
      "Data columns (total 32 columns):\n",
      " #   Column                    Non-Null Count  Dtype  \n",
      "---  ------                    --------------  -----  \n",
      " 0   id                        16096 non-null  object \n",
      " 1   activity_new              6551 non-null   object \n",
      " 2   campaign_disc_ele         0 non-null      float64\n",
      " 3   channel_sales             11878 non-null  object \n",
      " 4   cons_12m                  16096 non-null  int64  \n",
      " 5   cons_gas_12m              16096 non-null  int64  \n",
      " 6   cons_last_month           16096 non-null  int64  \n",
      " 7   date_activ                16096 non-null  object \n",
      " 8   date_end                  16094 non-null  object \n",
      " 9   date_first_activ          3508 non-null   object \n",
      " 10  date_modif_prod           15939 non-null  object \n",
      " 11  date_renewal              16056 non-null  object \n",
      " 12  forecast_base_bill_ele    3508 non-null   float64\n",
      " 13  forecast_base_bill_year   3508 non-null   float64\n",
      " 14  forecast_bill_12m         3508 non-null   float64\n",
      " 15  forecast_cons             3508 non-null   float64\n",
      " 16  forecast_cons_12m         16096 non-null  float64\n",
      " 17  forecast_cons_year        16096 non-null  int64  \n",
      " 18  forecast_discount_energy  15970 non-null  float64\n",
      " 19  forecast_meter_rent_12m   16096 non-null  float64\n",
      " 20  forecast_price_energy_p1  15970 non-null  float64\n",
      " 21  forecast_price_energy_p2  15970 non-null  float64\n",
      " 22  forecast_price_pow_p1     15970 non-null  float64\n",
      " 23  has_gas                   16096 non-null  object \n",
      " 24  imp_cons                  16096 non-null  float64\n",
      " 25  margin_gross_pow_ele      16083 non-null  float64\n",
      " 26  margin_net_pow_ele        16083 non-null  float64\n",
      " 27  nb_prod_act               16096 non-null  int64  \n",
      " 28  net_margin                16081 non-null  float64\n",
      " 29  num_years_antig           16096 non-null  int64  \n",
      " 30  origin_up                 16009 non-null  object \n",
      " 31  pow_max                   16093 non-null  float64\n",
      "dtypes: float64(16), int64(6), object(10)\n",
      "memory usage: 3.9+ MB\n"
     ]
    }
   ],
   "source": [
    "customer_data.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Customer data has missing values for some fields as the non-null count is lower in some fields. Also, the dataset has both numeric and non-numeric fields. Data types conversion would be required for fields containing dates."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4>Data Type Exploration & Conversion </h4>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "id                                  object\n",
       "activity_new                        object\n",
       "campaign_disc_ele                  float64\n",
       "channel_sales                       object\n",
       "cons_12m                             int64\n",
       "cons_gas_12m                         int64\n",
       "cons_last_month                      int64\n",
       "date_activ                  datetime64[ns]\n",
       "date_end                    datetime64[ns]\n",
       "date_first_activ            datetime64[ns]\n",
       "date_modif_prod             datetime64[ns]\n",
       "date_renewal                datetime64[ns]\n",
       "forecast_base_bill_ele             float64\n",
       "forecast_base_bill_year            float64\n",
       "forecast_bill_12m                  float64\n",
       "forecast_cons                      float64\n",
       "forecast_cons_12m                  float64\n",
       "forecast_cons_year                   int64\n",
       "forecast_discount_energy           float64\n",
       "forecast_meter_rent_12m            float64\n",
       "forecast_price_energy_p1           float64\n",
       "forecast_price_energy_p2           float64\n",
       "forecast_price_pow_p1              float64\n",
       "has_gas                             object\n",
       "imp_cons                           float64\n",
       "margin_gross_pow_ele               float64\n",
       "margin_net_pow_ele                 float64\n",
       "nb_prod_act                          int64\n",
       "net_margin                         float64\n",
       "num_years_antig                      int64\n",
       "origin_up                           object\n",
       "pow_max                            float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer_data['date_activ'] = pd.to_datetime(customer_data['date_activ'], format='%Y-%m-%d')\n",
    "customer_data['date_end'] = pd.to_datetime(customer_data['date_end'], format='%Y-%m-%d')\n",
    "customer_data['date_first_activ'] = pd.to_datetime(customer_data['date_first_activ'], format='%Y-%m-%d')\n",
    "customer_data['date_modif_prod'] = pd.to_datetime(customer_data['date_modif_prod'], format='%Y-%m-%d')\n",
    "customer_data['date_renewal'] = pd.to_datetime(customer_data['date_renewal'], format='%Y-%m-%d')\n",
    "\n",
    "customer_data.dtypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Exploring the fields with object datatype to see if those are categorical data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "apdekpcbwosbxepsfxclislboipuxpop    1577\n",
       "kkklcdamwfafdcfwofuscwfwadblfmce     422\n",
       "kwuslieomapmswolewpobpplkaooaaew     230\n",
       "fmwdwsxillemwbbwelxsampiuwwpcdcb     219\n",
       "ckfxocssowaeipxueikxcmaxdmcduxsa     189\n",
       "                                    ... \n",
       "bcsfemospxbiwoudpemmseeckfcpwfwu       1\n",
       "ikiucmkuisupefxcxfxxulkpwssppfuo       1\n",
       "sadbemboabpaxoesiucxoseffukxwsma       1\n",
       "dwdflbsopucwoxdmccmulwiiefiiabel       1\n",
       "upssicikedpwsfusuofwdxiopiuluubp       1\n",
       "Name: activity_new, Length: 419, dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer_data['activity_new'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are 419 different values for 'activity_new' field. It represents the category of the company's activity. However, 419 is a large number of categories. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "foosdfpfkusacimwkcsosbicdxkicaua    7377\n",
       "lmkebamcaaclubfxadlmueccxoimlema    2073\n",
       "usilxuppasemubllopkaafesmlibmsdf    1444\n",
       "ewpakwlliwisiwduibdlfmalxowmwpci     966\n",
       "sddiedcslfslkckwlfkdpoeeailfpeds      12\n",
       "epumfxlbckeskwekxbiuasklxalciiuu       4\n",
       "fixdbufsefwooaasfcxdxadsiekoceaa       2\n",
       "Name: channel_sales, dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer_data['channel_sales'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "f    13132\n",
       "t     2964\n",
       "Name: has_gas, dtype: int64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer_data['has_gas'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "lxidpiddsbxsbosboudacockeimpuepw    7825\n",
       "kamkkxfxxuwbdslkwifmmcsiusiuosws    4517\n",
       "ldkssxwpmemidmecebumciepifcamkci    3664\n",
       "usapbepcfoloekilkwsdiboslwaxobdp       2\n",
       "ewxeelcelemmiwuafmddpobolfuxioce       1\n",
       "Name: origin_up, dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer_data['origin_up'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4>Data Statistics</h4>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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_disc_ele</th>\n",
       "      <th>cons_12m</th>\n",
       "      <th>cons_gas_12m</th>\n",
       "      <th>cons_last_month</th>\n",
       "      <th>forecast_base_bill_ele</th>\n",
       "      <th>forecast_base_bill_year</th>\n",
       "      <th>forecast_bill_12m</th>\n",
       "      <th>forecast_cons</th>\n",
       "      <th>forecast_cons_12m</th>\n",
       "      <th>forecast_cons_year</th>\n",
       "      <th>...</th>\n",
       "      <th>forecast_price_energy_p1</th>\n",
       "      <th>forecast_price_energy_p2</th>\n",
       "      <th>forecast_price_pow_p1</th>\n",
       "      <th>imp_cons</th>\n",
       "      <th>margin_gross_pow_ele</th>\n",
       "      <th>margin_net_pow_ele</th>\n",
       "      <th>nb_prod_act</th>\n",
       "      <th>net_margin</th>\n",
       "      <th>num_years_antig</th>\n",
       "      <th>pow_max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1.609600e+04</td>\n",
       "      <td>1.609600e+04</td>\n",
       "      <td>1.609600e+04</td>\n",
       "      <td>3508.000000</td>\n",
       "      <td>3508.000000</td>\n",
       "      <td>3508.000000</td>\n",
       "      <td>3508.000000</td>\n",
       "      <td>16096.000000</td>\n",
       "      <td>16096.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>15970.000000</td>\n",
       "      <td>15970.000000</td>\n",
       "      <td>15970.000000</td>\n",
       "      <td>16096.000000</td>\n",
       "      <td>16083.000000</td>\n",
       "      <td>16083.000000</td>\n",
       "      <td>16096.000000</td>\n",
       "      <td>16081.000000</td>\n",
       "      <td>16096.000000</td>\n",
       "      <td>16093.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>NaN</td>\n",
       "      <td>1.948044e+05</td>\n",
       "      <td>3.191164e+04</td>\n",
       "      <td>1.946154e+04</td>\n",
       "      <td>335.843857</td>\n",
       "      <td>335.843857</td>\n",
       "      <td>3837.441866</td>\n",
       "      <td>206.845165</td>\n",
       "      <td>2370.555949</td>\n",
       "      <td>1907.347229</td>\n",
       "      <td>...</td>\n",
       "      <td>0.135901</td>\n",
       "      <td>0.052951</td>\n",
       "      <td>43.533496</td>\n",
       "      <td>196.123447</td>\n",
       "      <td>22.462276</td>\n",
       "      <td>21.460318</td>\n",
       "      <td>1.347788</td>\n",
       "      <td>217.987028</td>\n",
       "      <td>5.030629</td>\n",
       "      <td>20.604131</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>NaN</td>\n",
       "      <td>6.795151e+05</td>\n",
       "      <td>1.775885e+05</td>\n",
       "      <td>8.235676e+04</td>\n",
       "      <td>649.406000</td>\n",
       "      <td>649.406000</td>\n",
       "      <td>5425.744327</td>\n",
       "      <td>455.634288</td>\n",
       "      <td>4035.085664</td>\n",
       "      <td>5257.364759</td>\n",
       "      <td>...</td>\n",
       "      <td>0.026252</td>\n",
       "      <td>0.048617</td>\n",
       "      <td>5.212252</td>\n",
       "      <td>494.366979</td>\n",
       "      <td>23.700883</td>\n",
       "      <td>27.917349</td>\n",
       "      <td>1.459808</td>\n",
       "      <td>366.742030</td>\n",
       "      <td>1.676101</td>\n",
       "      <td>21.772421</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-1.252760e+05</td>\n",
       "      <td>-3.037000e+03</td>\n",
       "      <td>-9.138600e+04</td>\n",
       "      <td>-364.940000</td>\n",
       "      <td>-364.940000</td>\n",
       "      <td>-2503.480000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-16689.260000</td>\n",
       "      <td>-85627.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.122184</td>\n",
       "      <td>-9038.210000</td>\n",
       "      <td>-525.540000</td>\n",
       "      <td>-615.660000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-4148.990000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5.906250e+03</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1158.175000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>513.230000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.115237</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>40.606701</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>11.960000</td>\n",
       "      <td>11.950000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>51.970000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>12.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>NaN</td>\n",
       "      <td>1.533250e+04</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>9.010000e+02</td>\n",
       "      <td>162.955000</td>\n",
       "      <td>162.955000</td>\n",
       "      <td>2187.230000</td>\n",
       "      <td>42.215000</td>\n",
       "      <td>1179.160000</td>\n",
       "      <td>378.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.142881</td>\n",
       "      <td>0.086163</td>\n",
       "      <td>44.311378</td>\n",
       "      <td>44.465000</td>\n",
       "      <td>21.090000</td>\n",
       "      <td>20.970000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>119.680000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>13.856000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5.022150e+04</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>4.127000e+03</td>\n",
       "      <td>396.185000</td>\n",
       "      <td>396.185000</td>\n",
       "      <td>4246.555000</td>\n",
       "      <td>228.117500</td>\n",
       "      <td>2692.077500</td>\n",
       "      <td>1994.250000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.146348</td>\n",
       "      <td>0.098837</td>\n",
       "      <td>44.311378</td>\n",
       "      <td>218.090000</td>\n",
       "      <td>29.640000</td>\n",
       "      <td>29.640000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>275.810000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>19.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>NaN</td>\n",
       "      <td>1.609711e+07</td>\n",
       "      <td>4.188440e+06</td>\n",
       "      <td>4.538720e+06</td>\n",
       "      <td>12566.080000</td>\n",
       "      <td>12566.080000</td>\n",
       "      <td>81122.630000</td>\n",
       "      <td>9682.890000</td>\n",
       "      <td>103801.930000</td>\n",
       "      <td>175375.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.273963</td>\n",
       "      <td>0.195975</td>\n",
       "      <td>59.444710</td>\n",
       "      <td>15042.790000</td>\n",
       "      <td>374.640000</td>\n",
       "      <td>374.640000</td>\n",
       "      <td>32.000000</td>\n",
       "      <td>24570.650000</td>\n",
       "      <td>16.000000</td>\n",
       "      <td>500.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       campaign_disc_ele      cons_12m  cons_gas_12m  cons_last_month  \\\n",
       "count                0.0  1.609600e+04  1.609600e+04     1.609600e+04   \n",
       "mean                 NaN  1.948044e+05  3.191164e+04     1.946154e+04   \n",
       "std                  NaN  6.795151e+05  1.775885e+05     8.235676e+04   \n",
       "min                  NaN -1.252760e+05 -3.037000e+03    -9.138600e+04   \n",
       "25%                  NaN  5.906250e+03  0.000000e+00     0.000000e+00   \n",
       "50%                  NaN  1.533250e+04  0.000000e+00     9.010000e+02   \n",
       "75%                  NaN  5.022150e+04  0.000000e+00     4.127000e+03   \n",
       "max                  NaN  1.609711e+07  4.188440e+06     4.538720e+06   \n",
       "\n",
       "       forecast_base_bill_ele  forecast_base_bill_year  forecast_bill_12m  \\\n",
       "count             3508.000000              3508.000000        3508.000000   \n",
       "mean               335.843857               335.843857        3837.441866   \n",
       "std                649.406000               649.406000        5425.744327   \n",
       "min               -364.940000              -364.940000       -2503.480000   \n",
       "25%                  0.000000                 0.000000        1158.175000   \n",
       "50%                162.955000               162.955000        2187.230000   \n",
       "75%                396.185000               396.185000        4246.555000   \n",
       "max              12566.080000             12566.080000       81122.630000   \n",
       "\n",
       "       forecast_cons  forecast_cons_12m  forecast_cons_year  ...  \\\n",
       "count    3508.000000       16096.000000        16096.000000  ...   \n",
       "mean      206.845165        2370.555949         1907.347229  ...   \n",
       "std       455.634288        4035.085664         5257.364759  ...   \n",
       "min         0.000000      -16689.260000       -85627.000000  ...   \n",
       "25%         0.000000         513.230000            0.000000  ...   \n",
       "50%        42.215000        1179.160000          378.000000  ...   \n",
       "75%       228.117500        2692.077500         1994.250000  ...   \n",
       "max      9682.890000      103801.930000       175375.000000  ...   \n",
       "\n",
       "       forecast_price_energy_p1  forecast_price_energy_p2  \\\n",
       "count              15970.000000              15970.000000   \n",
       "mean                   0.135901                  0.052951   \n",
       "std                    0.026252                  0.048617   \n",
       "min                    0.000000                  0.000000   \n",
       "25%                    0.115237                  0.000000   \n",
       "50%                    0.142881                  0.086163   \n",
       "75%                    0.146348                  0.098837   \n",
       "max                    0.273963                  0.195975   \n",
       "\n",
       "       forecast_price_pow_p1      imp_cons  margin_gross_pow_ele  \\\n",
       "count           15970.000000  16096.000000          16083.000000   \n",
       "mean               43.533496    196.123447             22.462276   \n",
       "std                 5.212252    494.366979             23.700883   \n",
       "min                -0.122184  -9038.210000           -525.540000   \n",
       "25%                40.606701      0.000000             11.960000   \n",
       "50%                44.311378     44.465000             21.090000   \n",
       "75%                44.311378    218.090000             29.640000   \n",
       "max                59.444710  15042.790000            374.640000   \n",
       "\n",
       "       margin_net_pow_ele   nb_prod_act    net_margin  num_years_antig  \\\n",
       "count        16083.000000  16096.000000  16081.000000     16096.000000   \n",
       "mean            21.460318      1.347788    217.987028         5.030629   \n",
       "std             27.917349      1.459808    366.742030         1.676101   \n",
       "min           -615.660000      1.000000  -4148.990000         1.000000   \n",
       "25%             11.950000      1.000000     51.970000         4.000000   \n",
       "50%             20.970000      1.000000    119.680000         5.000000   \n",
       "75%             29.640000      1.000000    275.810000         6.000000   \n",
       "max            374.640000     32.000000  24570.650000        16.000000   \n",
       "\n",
       "            pow_max  \n",
       "count  16093.000000  \n",
       "mean      20.604131  \n",
       "std       21.772421  \n",
       "min        1.000000  \n",
       "25%       12.500000  \n",
       "50%       13.856000  \n",
       "75%       19.800000  \n",
       "max      500.000000  \n",
       "\n",
       "[8 rows x 22 columns]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer_data.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We later see that campaign_disc_ele variable is empty."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4> Variable Distributions </h4>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 25 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline \n",
    "customer_data.hist(bins=40, figsize=(20,15))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It appears that in most numerical fields, the largest bin has values either zero or closer to zero. campaign_disc_ele field has no data. Many histograms are skewed to right. Each field has a different range of values. While fields like cons_12m has a large scale, nb_prod_act has a smaller scale. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4> Outliers </h4>\n",
    "Outliers can be observed from the above histograms. As data seems skewed for some variables, there might be possible outliers. Can also use boxplots to explore outliers further."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline \n",
    "#customer_data.boxplot( figsize=(20,15))\n",
    "customer_data.boxplot(column = ['cons_12m','cons_gas_12m','cons_last_month'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "These variables have few outliers. Since there are only few, we can remove these instances from the dataset later."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline \n",
    "#customer_data.boxplot( figsize=(20,15))\n",
    "customer_data.boxplot(column = ['forecast_base_bill_ele','forecast_base_bill_ele', 'forecast_base_bill_year','forecast_bill_12m','forecast_cons','forecast_cons_12m'], figsize=(20,10))\n",
    "plt.xticks(rotation=90)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Can explore other variable in this way using boxplots. Also, the basic statistics generated using the .describe() method earlier indicates outliers, especially comparing the max value with the values of 75% quartile.\n",
    "We earlier counted the number of instances from each category for the categorical variable. There some categories had 1 or 2 instances when other categories had higher number of instances. Those can also be recognized as outliers."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4>Duplicates</h4>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(16096, 31)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Exclude the id variable and find and drop the instances where all variable values are similar, ie, similar rows\n",
    "dup_removed = customer_data.drop('id', axis=1).drop_duplicates()\n",
    "dup_removed.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Number of rows have not changed, and it is similar to custmer_data. So no duplicate rows have existed."
   ]
  },
  {
   "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>id</th>\n",
       "      <th>activity_new</th>\n",
       "      <th>campaign_disc_ele</th>\n",
       "      <th>channel_sales</th>\n",
       "      <th>cons_12m</th>\n",
       "      <th>cons_gas_12m</th>\n",
       "      <th>cons_last_month</th>\n",
       "      <th>date_activ</th>\n",
       "      <th>date_end</th>\n",
       "      <th>date_first_activ</th>\n",
       "      <th>...</th>\n",
       "      <th>forecast_price_pow_p1</th>\n",
       "      <th>has_gas</th>\n",
       "      <th>imp_cons</th>\n",
       "      <th>margin_gross_pow_ele</th>\n",
       "      <th>margin_net_pow_ele</th>\n",
       "      <th>nb_prod_act</th>\n",
       "      <th>net_margin</th>\n",
       "      <th>num_years_antig</th>\n",
       "      <th>origin_up</th>\n",
       "      <th>pow_max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>0 rows × 32 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [id, activity_new, campaign_disc_ele, channel_sales, cons_12m, cons_gas_12m, cons_last_month, date_activ, date_end, date_first_activ, date_modif_prod, date_renewal, forecast_base_bill_ele, forecast_base_bill_year, forecast_bill_12m, forecast_cons, forecast_cons_12m, forecast_cons_year, forecast_discount_energy, forecast_meter_rent_12m, forecast_price_energy_p1, forecast_price_energy_p2, forecast_price_pow_p1, has_gas, imp_cons, margin_gross_pow_ele, margin_net_pow_ele, nb_prod_act, net_margin, num_years_antig, origin_up, pow_max]\n",
       "Index: []\n",
       "\n",
       "[0 rows x 32 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Anotherway to check duplicates\n",
    "customer_data[customer_data.duplicated()]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4>Multicollinearity</h4>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "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_disc_ele</th>\n",
       "      <th>cons_12m</th>\n",
       "      <th>cons_gas_12m</th>\n",
       "      <th>cons_last_month</th>\n",
       "      <th>forecast_base_bill_ele</th>\n",
       "      <th>forecast_base_bill_year</th>\n",
       "      <th>forecast_bill_12m</th>\n",
       "      <th>forecast_cons</th>\n",
       "      <th>forecast_cons_12m</th>\n",
       "      <th>forecast_cons_year</th>\n",
       "      <th>...</th>\n",
       "      <th>forecast_price_energy_p1</th>\n",
       "      <th>forecast_price_energy_p2</th>\n",
       "      <th>forecast_price_pow_p1</th>\n",
       "      <th>imp_cons</th>\n",
       "      <th>margin_gross_pow_ele</th>\n",
       "      <th>margin_net_pow_ele</th>\n",
       "      <th>nb_prod_act</th>\n",
       "      <th>net_margin</th>\n",
       "      <th>num_years_antig</th>\n",
       "      <th>pow_max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>campaign_disc_ele</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cons_12m</th>\n",
       "      <td>NaN</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.471233</td>\n",
       "      <td>0.919545</td>\n",
       "      <td>0.132991</td>\n",
       "      <td>0.132991</td>\n",
       "      <td>0.149023</td>\n",
       "      <td>0.133147</td>\n",
       "      <td>0.165168</td>\n",
       "      <td>0.139526</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.033546</td>\n",
       "      <td>0.146758</td>\n",
       "      <td>-0.025418</td>\n",
       "      <td>0.139353</td>\n",
       "      <td>-0.065500</td>\n",
       "      <td>-0.045779</td>\n",
       "      <td>0.308567</td>\n",
       "      <td>0.120491</td>\n",
       "      <td>0.008810</td>\n",
       "      <td>0.102423</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cons_gas_12m</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.471233</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.447209</td>\n",
       "      <td>0.085733</td>\n",
       "      <td>0.085733</td>\n",
       "      <td>0.083604</td>\n",
       "      <td>0.076854</td>\n",
       "      <td>0.059525</td>\n",
       "      <td>0.057619</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.022416</td>\n",
       "      <td>0.078456</td>\n",
       "      <td>-0.027193</td>\n",
       "      <td>0.060609</td>\n",
       "      <td>-0.016867</td>\n",
       "      <td>-0.008242</td>\n",
       "      <td>0.272005</td>\n",
       "      <td>0.058930</td>\n",
       "      <td>-0.008626</td>\n",
       "      <td>0.052365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cons_last_month</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.919545</td>\n",
       "      <td>0.447209</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.136207</td>\n",
       "      <td>0.136207</td>\n",
       "      <td>0.134066</td>\n",
       "      <td>0.136816</td>\n",
       "      <td>0.129574</td>\n",
       "      <td>0.151476</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.024242</td>\n",
       "      <td>0.123164</td>\n",
       "      <td>-0.020057</td>\n",
       "      <td>0.153861</td>\n",
       "      <td>-0.054114</td>\n",
       "      <td>-0.037696</td>\n",
       "      <td>0.350711</td>\n",
       "      <td>0.096424</td>\n",
       "      <td>0.004860</td>\n",
       "      <td>0.089565</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forecast_base_bill_ele</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.132991</td>\n",
       "      <td>0.085733</td>\n",
       "      <td>0.136207</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.794776</td>\n",
       "      <td>0.964402</td>\n",
       "      <td>0.750961</td>\n",
       "      <td>0.958303</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.223693</td>\n",
       "      <td>0.351376</td>\n",
       "      <td>0.106436</td>\n",
       "      <td>0.964402</td>\n",
       "      <td>-0.088898</td>\n",
       "      <td>-0.054790</td>\n",
       "      <td>0.051419</td>\n",
       "      <td>0.468836</td>\n",
       "      <td>0.021869</td>\n",
       "      <td>0.585426</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forecast_base_bill_year</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.132991</td>\n",
       "      <td>0.085733</td>\n",
       "      <td>0.136207</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.794776</td>\n",
       "      <td>0.964402</td>\n",
       "      <td>0.750961</td>\n",
       "      <td>0.958303</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.223693</td>\n",
       "      <td>0.351376</td>\n",
       "      <td>0.106436</td>\n",
       "      <td>0.964402</td>\n",
       "      <td>-0.088898</td>\n",
       "      <td>-0.054790</td>\n",
       "      <td>0.051419</td>\n",
       "      <td>0.468836</td>\n",
       "      <td>0.021869</td>\n",
       "      <td>0.585426</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forecast_bill_12m</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.149023</td>\n",
       "      <td>0.083604</td>\n",
       "      <td>0.134066</td>\n",
       "      <td>0.794776</td>\n",
       "      <td>0.794776</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.751430</td>\n",
       "      <td>0.970487</td>\n",
       "      <td>0.797272</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.258372</td>\n",
       "      <td>0.376494</td>\n",
       "      <td>0.147365</td>\n",
       "      <td>0.751430</td>\n",
       "      <td>-0.129388</td>\n",
       "      <td>-0.086826</td>\n",
       "      <td>0.050683</td>\n",
       "      <td>0.656937</td>\n",
       "      <td>0.029917</td>\n",
       "      <td>0.711502</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forecast_cons</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.133147</td>\n",
       "      <td>0.076854</td>\n",
       "      <td>0.136816</td>\n",
       "      <td>0.964402</td>\n",
       "      <td>0.964402</td>\n",
       "      <td>0.751430</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.758825</td>\n",
       "      <td>0.974419</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.166692</td>\n",
       "      <td>0.290593</td>\n",
       "      <td>0.074713</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.111186</td>\n",
       "      <td>-0.075863</td>\n",
       "      <td>0.055135</td>\n",
       "      <td>0.489346</td>\n",
       "      <td>0.006925</td>\n",
       "      <td>0.457566</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forecast_cons_12m</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.165168</td>\n",
       "      <td>0.059525</td>\n",
       "      <td>0.129574</td>\n",
       "      <td>0.750961</td>\n",
       "      <td>0.750961</td>\n",
       "      <td>0.970487</td>\n",
       "      <td>0.758825</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.746076</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.217315</td>\n",
       "      <td>0.245845</td>\n",
       "      <td>0.058169</td>\n",
       "      <td>0.725550</td>\n",
       "      <td>-0.184179</td>\n",
       "      <td>-0.141642</td>\n",
       "      <td>0.013283</td>\n",
       "      <td>0.768871</td>\n",
       "      <td>0.064431</td>\n",
       "      <td>0.583119</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forecast_cons_year</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.139526</td>\n",
       "      <td>0.057619</td>\n",
       "      <td>0.151476</td>\n",
       "      <td>0.958303</td>\n",
       "      <td>0.958303</td>\n",
       "      <td>0.797272</td>\n",
       "      <td>0.974419</td>\n",
       "      <td>0.746076</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.206041</td>\n",
       "      <td>0.225691</td>\n",
       "      <td>0.053678</td>\n",
       "      <td>0.981732</td>\n",
       "      <td>-0.139177</td>\n",
       "      <td>-0.106576</td>\n",
       "      <td>0.013811</td>\n",
       "      <td>0.537701</td>\n",
       "      <td>0.066105</td>\n",
       "      <td>0.442228</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forecast_discount_energy</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.043708</td>\n",
       "      <td>-0.014945</td>\n",
       "      <td>-0.037773</td>\n",
       "      <td>0.011804</td>\n",
       "      <td>0.011804</td>\n",
       "      <td>0.003654</td>\n",
       "      <td>0.027360</td>\n",
       "      <td>0.014923</td>\n",
       "      <td>-0.009000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.319202</td>\n",
       "      <td>0.049174</td>\n",
       "      <td>0.024477</td>\n",
       "      <td>0.011383</td>\n",
       "      <td>0.199609</td>\n",
       "      <td>0.151140</td>\n",
       "      <td>0.055162</td>\n",
       "      <td>0.013500</td>\n",
       "      <td>-0.071723</td>\n",
       "      <td>-0.022646</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forecast_meter_rent_12m</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.085996</td>\n",
       "      <td>0.040327</td>\n",
       "      <td>0.076066</td>\n",
       "      <td>0.425970</td>\n",
       "      <td>0.425970</td>\n",
       "      <td>0.487659</td>\n",
       "      <td>0.328157</td>\n",
       "      <td>0.390550</td>\n",
       "      <td>0.329201</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.558751</td>\n",
       "      <td>0.636761</td>\n",
       "      <td>0.013597</td>\n",
       "      <td>0.296259</td>\n",
       "      <td>-0.018957</td>\n",
       "      <td>0.000856</td>\n",
       "      <td>0.000050</td>\n",
       "      <td>0.336343</td>\n",
       "      <td>0.112271</td>\n",
       "      <td>0.600594</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forecast_price_energy_p1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.033546</td>\n",
       "      <td>-0.022416</td>\n",
       "      <td>-0.024242</td>\n",
       "      <td>-0.223693</td>\n",
       "      <td>-0.223693</td>\n",
       "      <td>-0.258372</td>\n",
       "      <td>-0.166692</td>\n",
       "      <td>-0.217315</td>\n",
       "      <td>-0.206041</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.364849</td>\n",
       "      <td>0.389218</td>\n",
       "      <td>-0.164657</td>\n",
       "      <td>0.184782</td>\n",
       "      <td>0.029119</td>\n",
       "      <td>0.025854</td>\n",
       "      <td>-0.185221</td>\n",
       "      <td>-0.199922</td>\n",
       "      <td>-0.352961</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forecast_price_energy_p2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.146758</td>\n",
       "      <td>0.078456</td>\n",
       "      <td>0.123164</td>\n",
       "      <td>0.351376</td>\n",
       "      <td>0.351376</td>\n",
       "      <td>0.376494</td>\n",
       "      <td>0.290593</td>\n",
       "      <td>0.245845</td>\n",
       "      <td>0.225691</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.364849</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.137244</td>\n",
       "      <td>0.211061</td>\n",
       "      <td>0.063421</td>\n",
       "      <td>0.074075</td>\n",
       "      <td>0.025949</td>\n",
       "      <td>0.251761</td>\n",
       "      <td>0.102997</td>\n",
       "      <td>0.339373</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forecast_price_pow_p1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.025418</td>\n",
       "      <td>-0.027193</td>\n",
       "      <td>-0.020057</td>\n",
       "      <td>0.106436</td>\n",
       "      <td>0.106436</td>\n",
       "      <td>0.147365</td>\n",
       "      <td>0.074713</td>\n",
       "      <td>0.058169</td>\n",
       "      <td>0.053678</td>\n",
       "      <td>...</td>\n",
       "      <td>0.389218</td>\n",
       "      <td>-0.137244</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.051517</td>\n",
       "      <td>-0.114822</td>\n",
       "      <td>-0.134192</td>\n",
       "      <td>-0.011416</td>\n",
       "      <td>-0.005513</td>\n",
       "      <td>-0.037951</td>\n",
       "      <td>0.052583</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>imp_cons</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.139353</td>\n",
       "      <td>0.060609</td>\n",
       "      <td>0.153861</td>\n",
       "      <td>0.964402</td>\n",
       "      <td>0.964402</td>\n",
       "      <td>0.751430</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.725550</td>\n",
       "      <td>0.981732</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.164657</td>\n",
       "      <td>0.211061</td>\n",
       "      <td>0.051517</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.122137</td>\n",
       "      <td>-0.092279</td>\n",
       "      <td>0.019056</td>\n",
       "      <td>0.536779</td>\n",
       "      <td>0.051019</td>\n",
       "      <td>0.407694</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>margin_gross_pow_ele</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.065500</td>\n",
       "      <td>-0.016867</td>\n",
       "      <td>-0.054114</td>\n",
       "      <td>-0.088898</td>\n",
       "      <td>-0.088898</td>\n",
       "      <td>-0.129388</td>\n",
       "      <td>-0.111186</td>\n",
       "      <td>-0.184179</td>\n",
       "      <td>-0.139177</td>\n",
       "      <td>...</td>\n",
       "      <td>0.184782</td>\n",
       "      <td>0.063421</td>\n",
       "      <td>-0.114822</td>\n",
       "      <td>-0.122137</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.766521</td>\n",
       "      <td>-0.043771</td>\n",
       "      <td>-0.098609</td>\n",
       "      <td>-0.081516</td>\n",
       "      <td>-0.013654</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>margin_net_pow_ele</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.045779</td>\n",
       "      <td>-0.008242</td>\n",
       "      <td>-0.037696</td>\n",
       "      <td>-0.054790</td>\n",
       "      <td>-0.054790</td>\n",
       "      <td>-0.086826</td>\n",
       "      <td>-0.075863</td>\n",
       "      <td>-0.141642</td>\n",
       "      <td>-0.106576</td>\n",
       "      <td>...</td>\n",
       "      <td>0.029119</td>\n",
       "      <td>0.074075</td>\n",
       "      <td>-0.134192</td>\n",
       "      <td>-0.092279</td>\n",
       "      <td>0.766521</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.032199</td>\n",
       "      <td>-0.086364</td>\n",
       "      <td>-0.037913</td>\n",
       "      <td>-0.001202</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nb_prod_act</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.308567</td>\n",
       "      <td>0.272005</td>\n",
       "      <td>0.350711</td>\n",
       "      <td>0.051419</td>\n",
       "      <td>0.051419</td>\n",
       "      <td>0.050683</td>\n",
       "      <td>0.055135</td>\n",
       "      <td>0.013283</td>\n",
       "      <td>0.013811</td>\n",
       "      <td>...</td>\n",
       "      <td>0.025854</td>\n",
       "      <td>0.025949</td>\n",
       "      <td>-0.011416</td>\n",
       "      <td>0.019056</td>\n",
       "      <td>-0.043771</td>\n",
       "      <td>-0.032199</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.004547</td>\n",
       "      <td>0.009384</td>\n",
       "      <td>0.018390</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>net_margin</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.120491</td>\n",
       "      <td>0.058930</td>\n",
       "      <td>0.096424</td>\n",
       "      <td>0.468836</td>\n",
       "      <td>0.468836</td>\n",
       "      <td>0.656937</td>\n",
       "      <td>0.489346</td>\n",
       "      <td>0.768871</td>\n",
       "      <td>0.537701</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.185221</td>\n",
       "      <td>0.251761</td>\n",
       "      <td>-0.005513</td>\n",
       "      <td>0.536779</td>\n",
       "      <td>-0.098609</td>\n",
       "      <td>-0.086364</td>\n",
       "      <td>0.004547</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.034960</td>\n",
       "      <td>0.457120</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_years_antig</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.008810</td>\n",
       "      <td>-0.008626</td>\n",
       "      <td>0.004860</td>\n",
       "      <td>0.021869</td>\n",
       "      <td>0.021869</td>\n",
       "      <td>0.029917</td>\n",
       "      <td>0.006925</td>\n",
       "      <td>0.064431</td>\n",
       "      <td>0.066105</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.199922</td>\n",
       "      <td>0.102997</td>\n",
       "      <td>-0.037951</td>\n",
       "      <td>0.051019</td>\n",
       "      <td>-0.081516</td>\n",
       "      <td>-0.037913</td>\n",
       "      <td>0.009384</td>\n",
       "      <td>0.034960</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.084789</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pow_max</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.102423</td>\n",
       "      <td>0.052365</td>\n",
       "      <td>0.089565</td>\n",
       "      <td>0.585426</td>\n",
       "      <td>0.585426</td>\n",
       "      <td>0.711502</td>\n",
       "      <td>0.457566</td>\n",
       "      <td>0.583119</td>\n",
       "      <td>0.442228</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.352961</td>\n",
       "      <td>0.339373</td>\n",
       "      <td>0.052583</td>\n",
       "      <td>0.407694</td>\n",
       "      <td>-0.013654</td>\n",
       "      <td>-0.001202</td>\n",
       "      <td>0.018390</td>\n",
       "      <td>0.457120</td>\n",
       "      <td>0.084789</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>22 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                          campaign_disc_ele  cons_12m  cons_gas_12m  \\\n",
       "campaign_disc_ele                       NaN       NaN           NaN   \n",
       "cons_12m                                NaN  1.000000      0.471233   \n",
       "cons_gas_12m                            NaN  0.471233      1.000000   \n",
       "cons_last_month                         NaN  0.919545      0.447209   \n",
       "forecast_base_bill_ele                  NaN  0.132991      0.085733   \n",
       "forecast_base_bill_year                 NaN  0.132991      0.085733   \n",
       "forecast_bill_12m                       NaN  0.149023      0.083604   \n",
       "forecast_cons                           NaN  0.133147      0.076854   \n",
       "forecast_cons_12m                       NaN  0.165168      0.059525   \n",
       "forecast_cons_year                      NaN  0.139526      0.057619   \n",
       "forecast_discount_energy                NaN -0.043708     -0.014945   \n",
       "forecast_meter_rent_12m                 NaN  0.085996      0.040327   \n",
       "forecast_price_energy_p1                NaN -0.033546     -0.022416   \n",
       "forecast_price_energy_p2                NaN  0.146758      0.078456   \n",
       "forecast_price_pow_p1                   NaN -0.025418     -0.027193   \n",
       "imp_cons                                NaN  0.139353      0.060609   \n",
       "margin_gross_pow_ele                    NaN -0.065500     -0.016867   \n",
       "margin_net_pow_ele                      NaN -0.045779     -0.008242   \n",
       "nb_prod_act                             NaN  0.308567      0.272005   \n",
       "net_margin                              NaN  0.120491      0.058930   \n",
       "num_years_antig                         NaN  0.008810     -0.008626   \n",
       "pow_max                                 NaN  0.102423      0.052365   \n",
       "\n",
       "                          cons_last_month  forecast_base_bill_ele  \\\n",
       "campaign_disc_ele                     NaN                     NaN   \n",
       "cons_12m                         0.919545                0.132991   \n",
       "cons_gas_12m                     0.447209                0.085733   \n",
       "cons_last_month                  1.000000                0.136207   \n",
       "forecast_base_bill_ele           0.136207                1.000000   \n",
       "forecast_base_bill_year          0.136207                1.000000   \n",
       "forecast_bill_12m                0.134066                0.794776   \n",
       "forecast_cons                    0.136816                0.964402   \n",
       "forecast_cons_12m                0.129574                0.750961   \n",
       "forecast_cons_year               0.151476                0.958303   \n",
       "forecast_discount_energy        -0.037773                0.011804   \n",
       "forecast_meter_rent_12m          0.076066                0.425970   \n",
       "forecast_price_energy_p1        -0.024242               -0.223693   \n",
       "forecast_price_energy_p2         0.123164                0.351376   \n",
       "forecast_price_pow_p1           -0.020057                0.106436   \n",
       "imp_cons                         0.153861                0.964402   \n",
       "margin_gross_pow_ele            -0.054114               -0.088898   \n",
       "margin_net_pow_ele              -0.037696               -0.054790   \n",
       "nb_prod_act                      0.350711                0.051419   \n",
       "net_margin                       0.096424                0.468836   \n",
       "num_years_antig                  0.004860                0.021869   \n",
       "pow_max                          0.089565                0.585426   \n",
       "\n",
       "                          forecast_base_bill_year  forecast_bill_12m  \\\n",
       "campaign_disc_ele                             NaN                NaN   \n",
       "cons_12m                                 0.132991           0.149023   \n",
       "cons_gas_12m                             0.085733           0.083604   \n",
       "cons_last_month                          0.136207           0.134066   \n",
       "forecast_base_bill_ele                   1.000000           0.794776   \n",
       "forecast_base_bill_year                  1.000000           0.794776   \n",
       "forecast_bill_12m                        0.794776           1.000000   \n",
       "forecast_cons                            0.964402           0.751430   \n",
       "forecast_cons_12m                        0.750961           0.970487   \n",
       "forecast_cons_year                       0.958303           0.797272   \n",
       "forecast_discount_energy                 0.011804           0.003654   \n",
       "forecast_meter_rent_12m                  0.425970           0.487659   \n",
       "forecast_price_energy_p1                -0.223693          -0.258372   \n",
       "forecast_price_energy_p2                 0.351376           0.376494   \n",
       "forecast_price_pow_p1                    0.106436           0.147365   \n",
       "imp_cons                                 0.964402           0.751430   \n",
       "margin_gross_pow_ele                    -0.088898          -0.129388   \n",
       "margin_net_pow_ele                      -0.054790          -0.086826   \n",
       "nb_prod_act                              0.051419           0.050683   \n",
       "net_margin                               0.468836           0.656937   \n",
       "num_years_antig                          0.021869           0.029917   \n",
       "pow_max                                  0.585426           0.711502   \n",
       "\n",
       "                          forecast_cons  forecast_cons_12m  \\\n",
       "campaign_disc_ele                   NaN                NaN   \n",
       "cons_12m                       0.133147           0.165168   \n",
       "cons_gas_12m                   0.076854           0.059525   \n",
       "cons_last_month                0.136816           0.129574   \n",
       "forecast_base_bill_ele         0.964402           0.750961   \n",
       "forecast_base_bill_year        0.964402           0.750961   \n",
       "forecast_bill_12m              0.751430           0.970487   \n",
       "forecast_cons                  1.000000           0.758825   \n",
       "forecast_cons_12m              0.758825           1.000000   \n",
       "forecast_cons_year             0.974419           0.746076   \n",
       "forecast_discount_energy       0.027360           0.014923   \n",
       "forecast_meter_rent_12m        0.328157           0.390550   \n",
       "forecast_price_energy_p1      -0.166692          -0.217315   \n",
       "forecast_price_energy_p2       0.290593           0.245845   \n",
       "forecast_price_pow_p1          0.074713           0.058169   \n",
       "imp_cons                       1.000000           0.725550   \n",
       "margin_gross_pow_ele          -0.111186          -0.184179   \n",
       "margin_net_pow_ele            -0.075863          -0.141642   \n",
       "nb_prod_act                    0.055135           0.013283   \n",
       "net_margin                     0.489346           0.768871   \n",
       "num_years_antig                0.006925           0.064431   \n",
       "pow_max                        0.457566           0.583119   \n",
       "\n",
       "                          forecast_cons_year  ...  forecast_price_energy_p1  \\\n",
       "campaign_disc_ele                        NaN  ...                       NaN   \n",
       "cons_12m                            0.139526  ...                 -0.033546   \n",
       "cons_gas_12m                        0.057619  ...                 -0.022416   \n",
       "cons_last_month                     0.151476  ...                 -0.024242   \n",
       "forecast_base_bill_ele              0.958303  ...                 -0.223693   \n",
       "forecast_base_bill_year             0.958303  ...                 -0.223693   \n",
       "forecast_bill_12m                   0.797272  ...                 -0.258372   \n",
       "forecast_cons                       0.974419  ...                 -0.166692   \n",
       "forecast_cons_12m                   0.746076  ...                 -0.217315   \n",
       "forecast_cons_year                  1.000000  ...                 -0.206041   \n",
       "forecast_discount_energy           -0.009000  ...                  0.319202   \n",
       "forecast_meter_rent_12m             0.329201  ...                 -0.558751   \n",
       "forecast_price_energy_p1           -0.206041  ...                  1.000000   \n",
       "forecast_price_energy_p2            0.225691  ...                 -0.364849   \n",
       "forecast_price_pow_p1               0.053678  ...                  0.389218   \n",
       "imp_cons                            0.981732  ...                 -0.164657   \n",
       "margin_gross_pow_ele               -0.139177  ...                  0.184782   \n",
       "margin_net_pow_ele                 -0.106576  ...                  0.029119   \n",
       "nb_prod_act                         0.013811  ...                  0.025854   \n",
       "net_margin                          0.537701  ...                 -0.185221   \n",
       "num_years_antig                     0.066105  ...                 -0.199922   \n",
       "pow_max                             0.442228  ...                 -0.352961   \n",
       "\n",
       "                          forecast_price_energy_p2  forecast_price_pow_p1  \\\n",
       "campaign_disc_ele                              NaN                    NaN   \n",
       "cons_12m                                  0.146758              -0.025418   \n",
       "cons_gas_12m                              0.078456              -0.027193   \n",
       "cons_last_month                           0.123164              -0.020057   \n",
       "forecast_base_bill_ele                    0.351376               0.106436   \n",
       "forecast_base_bill_year                   0.351376               0.106436   \n",
       "forecast_bill_12m                         0.376494               0.147365   \n",
       "forecast_cons                             0.290593               0.074713   \n",
       "forecast_cons_12m                         0.245845               0.058169   \n",
       "forecast_cons_year                        0.225691               0.053678   \n",
       "forecast_discount_energy                  0.049174               0.024477   \n",
       "forecast_meter_rent_12m                   0.636761               0.013597   \n",
       "forecast_price_energy_p1                 -0.364849               0.389218   \n",
       "forecast_price_energy_p2                  1.000000              -0.137244   \n",
       "forecast_price_pow_p1                    -0.137244               1.000000   \n",
       "imp_cons                                  0.211061               0.051517   \n",
       "margin_gross_pow_ele                      0.063421              -0.114822   \n",
       "margin_net_pow_ele                        0.074075              -0.134192   \n",
       "nb_prod_act                               0.025949              -0.011416   \n",
       "net_margin                                0.251761              -0.005513   \n",
       "num_years_antig                           0.102997              -0.037951   \n",
       "pow_max                                   0.339373               0.052583   \n",
       "\n",
       "                          imp_cons  margin_gross_pow_ele  margin_net_pow_ele  \\\n",
       "campaign_disc_ele              NaN                   NaN                 NaN   \n",
       "cons_12m                  0.139353             -0.065500           -0.045779   \n",
       "cons_gas_12m              0.060609             -0.016867           -0.008242   \n",
       "cons_last_month           0.153861             -0.054114           -0.037696   \n",
       "forecast_base_bill_ele    0.964402             -0.088898           -0.054790   \n",
       "forecast_base_bill_year   0.964402             -0.088898           -0.054790   \n",
       "forecast_bill_12m         0.751430             -0.129388           -0.086826   \n",
       "forecast_cons             1.000000             -0.111186           -0.075863   \n",
       "forecast_cons_12m         0.725550             -0.184179           -0.141642   \n",
       "forecast_cons_year        0.981732             -0.139177           -0.106576   \n",
       "forecast_discount_energy  0.011383              0.199609            0.151140   \n",
       "forecast_meter_rent_12m   0.296259             -0.018957            0.000856   \n",
       "forecast_price_energy_p1 -0.164657              0.184782            0.029119   \n",
       "forecast_price_energy_p2  0.211061              0.063421            0.074075   \n",
       "forecast_price_pow_p1     0.051517             -0.114822           -0.134192   \n",
       "imp_cons                  1.000000             -0.122137           -0.092279   \n",
       "margin_gross_pow_ele     -0.122137              1.000000            0.766521   \n",
       "margin_net_pow_ele       -0.092279              0.766521            1.000000   \n",
       "nb_prod_act               0.019056             -0.043771           -0.032199   \n",
       "net_margin                0.536779             -0.098609           -0.086364   \n",
       "num_years_antig           0.051019             -0.081516           -0.037913   \n",
       "pow_max                   0.407694             -0.013654           -0.001202   \n",
       "\n",
       "                          nb_prod_act  net_margin  num_years_antig   pow_max  \n",
       "campaign_disc_ele                 NaN         NaN              NaN       NaN  \n",
       "cons_12m                     0.308567    0.120491         0.008810  0.102423  \n",
       "cons_gas_12m                 0.272005    0.058930        -0.008626  0.052365  \n",
       "cons_last_month              0.350711    0.096424         0.004860  0.089565  \n",
       "forecast_base_bill_ele       0.051419    0.468836         0.021869  0.585426  \n",
       "forecast_base_bill_year      0.051419    0.468836         0.021869  0.585426  \n",
       "forecast_bill_12m            0.050683    0.656937         0.029917  0.711502  \n",
       "forecast_cons                0.055135    0.489346         0.006925  0.457566  \n",
       "forecast_cons_12m            0.013283    0.768871         0.064431  0.583119  \n",
       "forecast_cons_year           0.013811    0.537701         0.066105  0.442228  \n",
       "forecast_discount_energy     0.055162    0.013500        -0.071723 -0.022646  \n",
       "forecast_meter_rent_12m      0.000050    0.336343         0.112271  0.600594  \n",
       "forecast_price_energy_p1     0.025854   -0.185221        -0.199922 -0.352961  \n",
       "forecast_price_energy_p2     0.025949    0.251761         0.102997  0.339373  \n",
       "forecast_price_pow_p1       -0.011416   -0.005513        -0.037951  0.052583  \n",
       "imp_cons                     0.019056    0.536779         0.051019  0.407694  \n",
       "margin_gross_pow_ele        -0.043771   -0.098609        -0.081516 -0.013654  \n",
       "margin_net_pow_ele          -0.032199   -0.086364        -0.037913 -0.001202  \n",
       "nb_prod_act                  1.000000    0.004547         0.009384  0.018390  \n",
       "net_margin                   0.004547    1.000000         0.034960  0.457120  \n",
       "num_years_antig              0.009384    0.034960         1.000000  0.084789  \n",
       "pow_max                      0.018390    0.457120         0.084789  1.000000  \n",
       "\n",
       "[22 rows x 22 columns]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Pearson Correlation. Calculates pairwise correlation between pairs of variable excluding null values\n",
    "corr_df = customer_data.corr()\n",
    "corr_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It can be seen from the above matrix that some pairs of variables have a very high positive correlation(>=0.9): Eg: cons_last_month and cons_12m, forecast_base_bill_ele and forecast_base_bill_year, forecast_base_bill_ele and forecast_cons, forecast_base_bill_ele and forecast_cons_year, forecast_base_bill_ele and imp_cons, forecast_base_bill_year and forecast_cons, forecast_base_bill_year and forecast_cons_year, forecast_base_bill_year and imp_cons, \n",
    "forecast_bill_12m and forecast_cons_12m, forecast_cons and forecast_cons_year, forecast_cons and imp_cons,\n",
    "\n",
    "Some pairs of variables have a high positive correlation(=0.7): Eg: forecast_base_bill_ele and forecast_bill_12m, forecast_base_bill_ele and forecast_cons_12m, forecast_base_bill_year and forecast_bill_12m, forecast_base_bill_year and forecast_cons_12m, forecast_bill_12m and forecast_base_bill_ele, forecast_bill_12m and forecast_base_bill_year, forecast_bill_12m and forecast_cons, forecast_bill_12m and forecast_cons_year, forecast_bill_12m and imp_cons, forecast_bill_12m and pow_max, forecast_cons and forecast_cons_12m, \n",
    "*some of these fields have a lot of missing values. As we see later, the missing values belong to same rows.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Combining customer_data(possible predictor variables) with churn_status (response variable)"
   ]
  },
  {
   "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 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>activity_new</th>\n",
       "      <th>campaign_disc_ele</th>\n",
       "      <th>channel_sales</th>\n",
       "      <th>cons_12m</th>\n",
       "      <th>cons_gas_12m</th>\n",
       "      <th>cons_last_month</th>\n",
       "      <th>date_activ</th>\n",
       "      <th>date_end</th>\n",
       "      <th>date_first_activ</th>\n",
       "      <th>...</th>\n",
       "      <th>has_gas</th>\n",
       "      <th>imp_cons</th>\n",
       "      <th>margin_gross_pow_ele</th>\n",
       "      <th>margin_net_pow_ele</th>\n",
       "      <th>nb_prod_act</th>\n",
       "      <th>net_margin</th>\n",
       "      <th>num_years_antig</th>\n",
       "      <th>origin_up</th>\n",
       "      <th>pow_max</th>\n",
       "      <th>churn</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>48ada52261e7cf58715202705a0451c9</td>\n",
       "      <td>esoiiifxdlbkcsluxmfuacbdckommixw</td>\n",
       "      <td>NaN</td>\n",
       "      <td>lmkebamcaaclubfxadlmueccxoimlema</td>\n",
       "      <td>309275</td>\n",
       "      <td>0</td>\n",
       "      <td>10025</td>\n",
       "      <td>2012-11-07</td>\n",
       "      <td>2016-11-06</td>\n",
       "      <td>NaT</td>\n",
       "      <td>...</td>\n",
       "      <td>f</td>\n",
       "      <td>831.8</td>\n",
       "      <td>-41.76</td>\n",
       "      <td>-41.76</td>\n",
       "      <td>1</td>\n",
       "      <td>1732.36</td>\n",
       "      <td>3</td>\n",
       "      <td>ldkssxwpmemidmecebumciepifcamkci</td>\n",
       "      <td>180.000</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>24011ae4ebbe3035111d65fa7c15bc57</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>foosdfpfkusacimwkcsosbicdxkicaua</td>\n",
       "      <td>0</td>\n",
       "      <td>54946</td>\n",
       "      <td>0</td>\n",
       "      <td>2013-06-15</td>\n",
       "      <td>2016-06-15</td>\n",
       "      <td>NaT</td>\n",
       "      <td>...</td>\n",
       "      <td>t</td>\n",
       "      <td>0.0</td>\n",
       "      <td>25.44</td>\n",
       "      <td>25.44</td>\n",
       "      <td>2</td>\n",
       "      <td>678.99</td>\n",
       "      <td>3</td>\n",
       "      <td>lxidpiddsbxsbosboudacockeimpuepw</td>\n",
       "      <td>43.648</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>d29c2c54acc38ff3c0614d0a653813dd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4660</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2009-08-21</td>\n",
       "      <td>2016-08-30</td>\n",
       "      <td>NaT</td>\n",
       "      <td>...</td>\n",
       "      <td>f</td>\n",
       "      <td>0.0</td>\n",
       "      <td>16.38</td>\n",
       "      <td>16.38</td>\n",
       "      <td>1</td>\n",
       "      <td>18.89</td>\n",
       "      <td>6</td>\n",
       "      <td>kamkkxfxxuwbdslkwifmmcsiusiuosws</td>\n",
       "      <td>13.800</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>764c75f661154dac3a6c254cd082ea7d</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>foosdfpfkusacimwkcsosbicdxkicaua</td>\n",
       "      <td>544</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2010-04-16</td>\n",
       "      <td>2016-04-16</td>\n",
       "      <td>NaT</td>\n",
       "      <td>...</td>\n",
       "      <td>f</td>\n",
       "      <td>0.0</td>\n",
       "      <td>28.60</td>\n",
       "      <td>28.60</td>\n",
       "      <td>1</td>\n",
       "      <td>6.60</td>\n",
       "      <td>6</td>\n",
       "      <td>kamkkxfxxuwbdslkwifmmcsiusiuosws</td>\n",
       "      <td>13.856</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>bba03439a292a1e166f80264c16191cb</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>lmkebamcaaclubfxadlmueccxoimlema</td>\n",
       "      <td>1584</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2010-03-30</td>\n",
       "      <td>2016-03-30</td>\n",
       "      <td>NaT</td>\n",
       "      <td>...</td>\n",
       "      <td>f</td>\n",
       "      <td>0.0</td>\n",
       "      <td>30.22</td>\n",
       "      <td>30.22</td>\n",
       "      <td>1</td>\n",
       "      <td>25.46</td>\n",
       "      <td>6</td>\n",
       "      <td>kamkkxfxxuwbdslkwifmmcsiusiuosws</td>\n",
       "      <td>13.200</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 33 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 id                      activity_new  \\\n",
       "0  48ada52261e7cf58715202705a0451c9  esoiiifxdlbkcsluxmfuacbdckommixw   \n",
       "1  24011ae4ebbe3035111d65fa7c15bc57                               NaN   \n",
       "2  d29c2c54acc38ff3c0614d0a653813dd                               NaN   \n",
       "3  764c75f661154dac3a6c254cd082ea7d                               NaN   \n",
       "4  bba03439a292a1e166f80264c16191cb                               NaN   \n",
       "\n",
       "   campaign_disc_ele                     channel_sales  cons_12m  \\\n",
       "0                NaN  lmkebamcaaclubfxadlmueccxoimlema    309275   \n",
       "1                NaN  foosdfpfkusacimwkcsosbicdxkicaua         0   \n",
       "2                NaN                               NaN      4660   \n",
       "3                NaN  foosdfpfkusacimwkcsosbicdxkicaua       544   \n",
       "4                NaN  lmkebamcaaclubfxadlmueccxoimlema      1584   \n",
       "\n",
       "   cons_gas_12m  cons_last_month date_activ   date_end date_first_activ  ...  \\\n",
       "0             0            10025 2012-11-07 2016-11-06              NaT  ...   \n",
       "1         54946                0 2013-06-15 2016-06-15              NaT  ...   \n",
       "2             0                0 2009-08-21 2016-08-30              NaT  ...   \n",
       "3             0                0 2010-04-16 2016-04-16              NaT  ...   \n",
       "4             0                0 2010-03-30 2016-03-30              NaT  ...   \n",
       "\n",
       "  has_gas imp_cons  margin_gross_pow_ele  margin_net_pow_ele  nb_prod_act  \\\n",
       "0       f    831.8                -41.76              -41.76            1   \n",
       "1       t      0.0                 25.44               25.44            2   \n",
       "2       f      0.0                 16.38               16.38            1   \n",
       "3       f      0.0                 28.60               28.60            1   \n",
       "4       f      0.0                 30.22               30.22            1   \n",
       "\n",
       "   net_margin  num_years_antig                         origin_up  pow_max  \\\n",
       "0     1732.36                3  ldkssxwpmemidmecebumciepifcamkci  180.000   \n",
       "1      678.99                3  lxidpiddsbxsbosboudacockeimpuepw   43.648   \n",
       "2       18.89                6  kamkkxfxxuwbdslkwifmmcsiusiuosws   13.800   \n",
       "3        6.60                6  kamkkxfxxuwbdslkwifmmcsiusiuosws   13.856   \n",
       "4       25.46                6  kamkkxfxxuwbdslkwifmmcsiusiuosws   13.200   \n",
       "\n",
       "   churn  \n",
       "0      0  \n",
       "1      1  \n",
       "2      0  \n",
       "3      0  \n",
       "4      0  \n",
       "\n",
       "[5 rows x 33 columns]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged_customer_churn = pd.merge(customer_data, churn_status, on = \"id\", how = \"inner\")\n",
    "merged_customer_churn.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "churn                       1.000000\n",
       "margin_gross_pow_ele        0.080158\n",
       "margin_net_pow_ele          0.063187\n",
       "forecast_meter_rent_12m     0.029971\n",
       "net_margin                  0.029308\n",
       "forecast_price_energy_p2    0.025597\n",
       "forecast_discount_energy    0.012344\n",
       "pow_max                     0.009456\n",
       "forecast_cons_12m           0.007395\n",
       "forecast_bill_12m           0.006909\n",
       "forecast_price_pow_p1       0.004034\n",
       "imp_cons                    0.003417\n",
       "forecast_cons_year          0.002756\n",
       "forecast_base_bill_year     0.000433\n",
       "forecast_base_bill_ele      0.000433\n",
       "forecast_price_energy_p1   -0.003337\n",
       "forecast_cons              -0.005247\n",
       "nb_prod_act                -0.022609\n",
       "cons_gas_12m               -0.040880\n",
       "cons_last_month            -0.046931\n",
       "cons_12m                   -0.051759\n",
       "num_years_antig            -0.071565\n",
       "campaign_disc_ele                NaN\n",
       "Name: churn, dtype: float64"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Calculating correlation between churn variable and other predictor variables\n",
    "corr_df = merged_customer_churn.corr()\n",
    "corr_df['churn'].sort_values(ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There doesn't seem to be any high positive or high negative correlation between the churn values and other variables. \n",
    "\n",
    "Churn vs retention percentages in the dataset:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    90.090706\n",
       "1     9.909294\n",
       "Name: churn, dtype: float64"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged_customer_churn['churn'].value_counts()/len(merged_customer_churn.index)*100"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There is around 10% of churned customers and 90% of retention "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4>Missing Data</h4>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "colours = ['#0B6121','#DF013A'] \n",
    "sns.heatmap(customer_data.isnull(), cmap=sns.color_palette(colours),cbar=False)\n",
    "#sns.heatmap(customer_data.isnull(), cbar=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Red indicates missing data and green indicates non-missing data in this heatmap. It can be seen that some fields/columns have a large portions of missing data, especially, activity_new, campaign_disc_ele, channel_sales, date_first_activ, forecast_base_bill_ele, forecast_base_bill_year,forecast_bill_12m, forecast_cons. It can also be observed that there's a possible correlation in the missing data locations of the date_first_activ, forecast_base_bill_ele, forecast_base_bill_year,forecast_bill_12m and forecast_cons columns. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "id - 0.0%\n",
      "activity_new - 59.3%\n",
      "campaign_disc_ele - 100.0%\n",
      "channel_sales - 26.21%\n",
      "cons_12m - 0.0%\n",
      "cons_gas_12m - 0.0%\n",
      "cons_last_month - 0.0%\n",
      "date_activ - 0.0%\n",
      "date_end - 0.01%\n",
      "date_first_activ - 78.21%\n",
      "date_modif_prod - 0.98%\n",
      "date_renewal - 0.25%\n",
      "forecast_base_bill_ele - 78.21%\n",
      "forecast_base_bill_year - 78.21%\n",
      "forecast_bill_12m - 78.21%\n",
      "forecast_cons - 78.21%\n",
      "forecast_cons_12m - 0.0%\n",
      "forecast_cons_year - 0.0%\n",
      "forecast_discount_energy - 0.78%\n",
      "forecast_meter_rent_12m - 0.0%\n",
      "forecast_price_energy_p1 - 0.78%\n",
      "forecast_price_energy_p2 - 0.78%\n",
      "forecast_price_pow_p1 - 0.78%\n",
      "has_gas - 0.0%\n",
      "imp_cons - 0.0%\n",
      "margin_gross_pow_ele - 0.08%\n",
      "margin_net_pow_ele - 0.08%\n",
      "nb_prod_act - 0.0%\n",
      "net_margin - 0.09%\n",
      "num_years_antig - 0.0%\n",
      "origin_up - 0.54%\n",
      "pow_max - 0.02%\n"
     ]
    }
   ],
   "source": [
    "for col in customer_data.columns:\n",
    "    missing_cells = customer_data[col].isnull().sum()\n",
    "    print( '{} - {}%'.format(col, round((missing_cells/customer_data.shape[0])*100,2)))  \n",
    "    #print( '{} - {}'.format(col, missing_cells))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To handle the missing data, we can consider either to drop the features or to impute values. campaign_disc_ele feature can be dropped as it's missing data percentage is 100%. \n",
    "Also, date_first_activ, forecast_base_bill_ele, forecast_base_bill_year,forecast_bill_12m, forecast_cons all has 78% of missing data. As it is a higher percentage those columns can be dropped. Since it was earlier observed from the heatmap that the missing values of these columns belong to similar rows, it can be considered to drop the rows instead of the columns. However, since 78% is a high percentage, dropping the rows will substantially reduce the number of instances in the dataset. Hence, columns are dropped. \n",
    "For other columns with missing data, data imputation strategies are explored.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "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>activity_new</th>\n",
       "      <th>channel_sales</th>\n",
       "      <th>cons_12m</th>\n",
       "      <th>cons_gas_12m</th>\n",
       "      <th>cons_last_month</th>\n",
       "      <th>date_activ</th>\n",
       "      <th>date_end</th>\n",
       "      <th>date_modif_prod</th>\n",
       "      <th>date_renewal</th>\n",
       "      <th>...</th>\n",
       "      <th>forecast_price_pow_p1</th>\n",
       "      <th>has_gas</th>\n",
       "      <th>imp_cons</th>\n",
       "      <th>margin_gross_pow_ele</th>\n",
       "      <th>margin_net_pow_ele</th>\n",
       "      <th>nb_prod_act</th>\n",
       "      <th>net_margin</th>\n",
       "      <th>num_years_antig</th>\n",
       "      <th>origin_up</th>\n",
       "      <th>pow_max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>48ada52261e7cf58715202705a0451c9</td>\n",
       "      <td>esoiiifxdlbkcsluxmfuacbdckommixw</td>\n",
       "      <td>lmkebamcaaclubfxadlmueccxoimlema</td>\n",
       "      <td>309275</td>\n",
       "      <td>0</td>\n",
       "      <td>10025</td>\n",
       "      <td>2012-11-07</td>\n",
       "      <td>2016-11-06</td>\n",
       "      <td>2012-11-07</td>\n",
       "      <td>2015-11-09</td>\n",
       "      <td>...</td>\n",
       "      <td>58.995952</td>\n",
       "      <td>f</td>\n",
       "      <td>831.80</td>\n",
       "      <td>-41.76</td>\n",
       "      <td>-41.76</td>\n",
       "      <td>1</td>\n",
       "      <td>1732.36</td>\n",
       "      <td>3</td>\n",
       "      <td>ldkssxwpmemidmecebumciepifcamkci</td>\n",
       "      <td>180.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>24011ae4ebbe3035111d65fa7c15bc57</td>\n",
       "      <td>NaN</td>\n",
       "      <td>foosdfpfkusacimwkcsosbicdxkicaua</td>\n",
       "      <td>0</td>\n",
       "      <td>54946</td>\n",
       "      <td>0</td>\n",
       "      <td>2013-06-15</td>\n",
       "      <td>2016-06-15</td>\n",
       "      <td>NaT</td>\n",
       "      <td>2015-06-23</td>\n",
       "      <td>...</td>\n",
       "      <td>40.606701</td>\n",
       "      <td>t</td>\n",
       "      <td>0.00</td>\n",
       "      <td>25.44</td>\n",
       "      <td>25.44</td>\n",
       "      <td>2</td>\n",
       "      <td>678.99</td>\n",
       "      <td>3</td>\n",
       "      <td>lxidpiddsbxsbosboudacockeimpuepw</td>\n",
       "      <td>43.648</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>d29c2c54acc38ff3c0614d0a653813dd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4660</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2009-08-21</td>\n",
       "      <td>2016-08-30</td>\n",
       "      <td>2009-08-21</td>\n",
       "      <td>2015-08-31</td>\n",
       "      <td>...</td>\n",
       "      <td>44.311378</td>\n",
       "      <td>f</td>\n",
       "      <td>0.00</td>\n",
       "      <td>16.38</td>\n",
       "      <td>16.38</td>\n",
       "      <td>1</td>\n",
       "      <td>18.89</td>\n",
       "      <td>6</td>\n",
       "      <td>kamkkxfxxuwbdslkwifmmcsiusiuosws</td>\n",
       "      <td>13.800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>764c75f661154dac3a6c254cd082ea7d</td>\n",
       "      <td>NaN</td>\n",
       "      <td>foosdfpfkusacimwkcsosbicdxkicaua</td>\n",
       "      <td>544</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2010-04-16</td>\n",
       "      <td>2016-04-16</td>\n",
       "      <td>2010-04-16</td>\n",
       "      <td>2015-04-17</td>\n",
       "      <td>...</td>\n",
       "      <td>44.311378</td>\n",
       "      <td>f</td>\n",
       "      <td>0.00</td>\n",
       "      <td>28.60</td>\n",
       "      <td>28.60</td>\n",
       "      <td>1</td>\n",
       "      <td>6.60</td>\n",
       "      <td>6</td>\n",
       "      <td>kamkkxfxxuwbdslkwifmmcsiusiuosws</td>\n",
       "      <td>13.856</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>bba03439a292a1e166f80264c16191cb</td>\n",
       "      <td>NaN</td>\n",
       "      <td>lmkebamcaaclubfxadlmueccxoimlema</td>\n",
       "      <td>1584</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2010-03-30</td>\n",
       "      <td>2016-03-30</td>\n",
       "      <td>2010-03-30</td>\n",
       "      <td>2015-03-31</td>\n",
       "      <td>...</td>\n",
       "      <td>44.311378</td>\n",
       "      <td>f</td>\n",
       "      <td>0.00</td>\n",
       "      <td>30.22</td>\n",
       "      <td>30.22</td>\n",
       "      <td>1</td>\n",
       "      <td>25.46</td>\n",
       "      <td>6</td>\n",
       "      <td>kamkkxfxxuwbdslkwifmmcsiusiuosws</td>\n",
       "      <td>13.200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16091</th>\n",
       "      <td>18463073fb097fc0ac5d3e040f356987</td>\n",
       "      <td>NaN</td>\n",
       "      <td>foosdfpfkusacimwkcsosbicdxkicaua</td>\n",
       "      <td>32270</td>\n",
       "      <td>47940</td>\n",
       "      <td>0</td>\n",
       "      <td>2012-05-24</td>\n",
       "      <td>2016-05-08</td>\n",
       "      <td>2015-05-08</td>\n",
       "      <td>2014-05-26</td>\n",
       "      <td>...</td>\n",
       "      <td>44.311378</td>\n",
       "      <td>t</td>\n",
       "      <td>0.00</td>\n",
       "      <td>27.88</td>\n",
       "      <td>27.88</td>\n",
       "      <td>2</td>\n",
       "      <td>381.77</td>\n",
       "      <td>4</td>\n",
       "      <td>lxidpiddsbxsbosboudacockeimpuepw</td>\n",
       "      <td>15.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16092</th>\n",
       "      <td>d0a6f71671571ed83b2645d23af6de00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>foosdfpfkusacimwkcsosbicdxkicaua</td>\n",
       "      <td>7223</td>\n",
       "      <td>0</td>\n",
       "      <td>181</td>\n",
       "      <td>2012-08-27</td>\n",
       "      <td>2016-08-27</td>\n",
       "      <td>2012-08-27</td>\n",
       "      <td>2015-08-28</td>\n",
       "      <td>...</td>\n",
       "      <td>58.995952</td>\n",
       "      <td>f</td>\n",
       "      <td>15.94</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>1</td>\n",
       "      <td>90.34</td>\n",
       "      <td>3</td>\n",
       "      <td>lxidpiddsbxsbosboudacockeimpuepw</td>\n",
       "      <td>6.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16093</th>\n",
       "      <td>10e6828ddd62cbcf687cb74928c4c2d2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>foosdfpfkusacimwkcsosbicdxkicaua</td>\n",
       "      <td>1844</td>\n",
       "      <td>0</td>\n",
       "      <td>179</td>\n",
       "      <td>2012-02-08</td>\n",
       "      <td>2016-02-07</td>\n",
       "      <td>2012-02-08</td>\n",
       "      <td>2015-02-09</td>\n",
       "      <td>...</td>\n",
       "      <td>40.606701</td>\n",
       "      <td>f</td>\n",
       "      <td>18.05</td>\n",
       "      <td>39.84</td>\n",
       "      <td>39.84</td>\n",
       "      <td>1</td>\n",
       "      <td>20.38</td>\n",
       "      <td>4</td>\n",
       "      <td>lxidpiddsbxsbosboudacockeimpuepw</td>\n",
       "      <td>15.935</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16094</th>\n",
       "      <td>1cf20fd6206d7678d5bcafd28c53b4db</td>\n",
       "      <td>NaN</td>\n",
       "      <td>foosdfpfkusacimwkcsosbicdxkicaua</td>\n",
       "      <td>131</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2012-08-30</td>\n",
       "      <td>2016-08-30</td>\n",
       "      <td>2012-08-30</td>\n",
       "      <td>2015-08-31</td>\n",
       "      <td>...</td>\n",
       "      <td>44.311378</td>\n",
       "      <td>f</td>\n",
       "      <td>0.00</td>\n",
       "      <td>13.08</td>\n",
       "      <td>13.08</td>\n",
       "      <td>1</td>\n",
       "      <td>0.96</td>\n",
       "      <td>3</td>\n",
       "      <td>lxidpiddsbxsbosboudacockeimpuepw</td>\n",
       "      <td>11.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16095</th>\n",
       "      <td>563dde550fd624d7352f3de77c0cdfcd</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>8730</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2009-12-18</td>\n",
       "      <td>2016-12-17</td>\n",
       "      <td>2009-12-18</td>\n",
       "      <td>2015-12-21</td>\n",
       "      <td>...</td>\n",
       "      <td>45.311378</td>\n",
       "      <td>f</td>\n",
       "      <td>0.00</td>\n",
       "      <td>11.84</td>\n",
       "      <td>11.84</td>\n",
       "      <td>1</td>\n",
       "      <td>96.34</td>\n",
       "      <td>6</td>\n",
       "      <td>ldkssxwpmemidmecebumciepifcamkci</td>\n",
       "      <td>10.392</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>16096 rows × 26 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     id                      activity_new  \\\n",
       "0      48ada52261e7cf58715202705a0451c9  esoiiifxdlbkcsluxmfuacbdckommixw   \n",
       "1      24011ae4ebbe3035111d65fa7c15bc57                               NaN   \n",
       "2      d29c2c54acc38ff3c0614d0a653813dd                               NaN   \n",
       "3      764c75f661154dac3a6c254cd082ea7d                               NaN   \n",
       "4      bba03439a292a1e166f80264c16191cb                               NaN   \n",
       "...                                 ...                               ...   \n",
       "16091  18463073fb097fc0ac5d3e040f356987                               NaN   \n",
       "16092  d0a6f71671571ed83b2645d23af6de00                               NaN   \n",
       "16093  10e6828ddd62cbcf687cb74928c4c2d2                               NaN   \n",
       "16094  1cf20fd6206d7678d5bcafd28c53b4db                               NaN   \n",
       "16095  563dde550fd624d7352f3de77c0cdfcd                               NaN   \n",
       "\n",
       "                          channel_sales  cons_12m  cons_gas_12m  \\\n",
       "0      lmkebamcaaclubfxadlmueccxoimlema    309275             0   \n",
       "1      foosdfpfkusacimwkcsosbicdxkicaua         0         54946   \n",
       "2                                   NaN      4660             0   \n",
       "3      foosdfpfkusacimwkcsosbicdxkicaua       544             0   \n",
       "4      lmkebamcaaclubfxadlmueccxoimlema      1584             0   \n",
       "...                                 ...       ...           ...   \n",
       "16091  foosdfpfkusacimwkcsosbicdxkicaua     32270         47940   \n",
       "16092  foosdfpfkusacimwkcsosbicdxkicaua      7223             0   \n",
       "16093  foosdfpfkusacimwkcsosbicdxkicaua      1844             0   \n",
       "16094  foosdfpfkusacimwkcsosbicdxkicaua       131             0   \n",
       "16095                               NaN      8730             0   \n",
       "\n",
       "       cons_last_month date_activ   date_end date_modif_prod date_renewal  \\\n",
       "0                10025 2012-11-07 2016-11-06      2012-11-07   2015-11-09   \n",
       "1                    0 2013-06-15 2016-06-15             NaT   2015-06-23   \n",
       "2                    0 2009-08-21 2016-08-30      2009-08-21   2015-08-31   \n",
       "3                    0 2010-04-16 2016-04-16      2010-04-16   2015-04-17   \n",
       "4                    0 2010-03-30 2016-03-30      2010-03-30   2015-03-31   \n",
       "...                ...        ...        ...             ...          ...   \n",
       "16091                0 2012-05-24 2016-05-08      2015-05-08   2014-05-26   \n",
       "16092              181 2012-08-27 2016-08-27      2012-08-27   2015-08-28   \n",
       "16093              179 2012-02-08 2016-02-07      2012-02-08   2015-02-09   \n",
       "16094                0 2012-08-30 2016-08-30      2012-08-30   2015-08-31   \n",
       "16095                0 2009-12-18 2016-12-17      2009-12-18   2015-12-21   \n",
       "\n",
       "       ...  forecast_price_pow_p1  has_gas  imp_cons  margin_gross_pow_ele  \\\n",
       "0      ...              58.995952        f    831.80                -41.76   \n",
       "1      ...              40.606701        t      0.00                 25.44   \n",
       "2      ...              44.311378        f      0.00                 16.38   \n",
       "3      ...              44.311378        f      0.00                 28.60   \n",
       "4      ...              44.311378        f      0.00                 30.22   \n",
       "...    ...                    ...      ...       ...                   ...   \n",
       "16091  ...              44.311378        t      0.00                 27.88   \n",
       "16092  ...              58.995952        f     15.94                  0.00   \n",
       "16093  ...              40.606701        f     18.05                 39.84   \n",
       "16094  ...              44.311378        f      0.00                 13.08   \n",
       "16095  ...              45.311378        f      0.00                 11.84   \n",
       "\n",
       "       margin_net_pow_ele  nb_prod_act  net_margin num_years_antig  \\\n",
       "0                  -41.76            1     1732.36               3   \n",
       "1                   25.44            2      678.99               3   \n",
       "2                   16.38            1       18.89               6   \n",
       "3                   28.60            1        6.60               6   \n",
       "4                   30.22            1       25.46               6   \n",
       "...                   ...          ...         ...             ...   \n",
       "16091               27.88            2      381.77               4   \n",
       "16092                0.00            1       90.34               3   \n",
       "16093               39.84            1       20.38               4   \n",
       "16094               13.08            1        0.96               3   \n",
       "16095               11.84            1       96.34               6   \n",
       "\n",
       "                              origin_up  pow_max  \n",
       "0      ldkssxwpmemidmecebumciepifcamkci  180.000  \n",
       "1      lxidpiddsbxsbosboudacockeimpuepw   43.648  \n",
       "2      kamkkxfxxuwbdslkwifmmcsiusiuosws   13.800  \n",
       "3      kamkkxfxxuwbdslkwifmmcsiusiuosws   13.856  \n",
       "4      kamkkxfxxuwbdslkwifmmcsiusiuosws   13.200  \n",
       "...                                 ...      ...  \n",
       "16091  lxidpiddsbxsbosboudacockeimpuepw   15.000  \n",
       "16092  lxidpiddsbxsbosboudacockeimpuepw    6.000  \n",
       "16093  lxidpiddsbxsbosboudacockeimpuepw   15.935  \n",
       "16094  lxidpiddsbxsbosboudacockeimpuepw   11.000  \n",
       "16095  ldkssxwpmemidmecebumciepifcamkci   10.392  \n",
       "\n",
       "[16096 rows x 26 columns]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer_data.drop(columns = ['campaign_disc_ele', 'date_first_activ', 'forecast_base_bill_ele', 'forecast_base_bill_year','forecast_bill_12m', 'forecast_cons'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "activity_new : This variable represents the category of the company's activity. Hence, missing data are filled with 'miscellaneous' to indicate that no category is assigned.\n",
    "channel_sales: This variable represents the code of the sales channel. It is also categorical. Missing data are filled with 'noinfo' to indicate that no category is assigned."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "customer_data['activity_new'].fillna('miscellaneous',inplace=True) \n",
    "customer_data['channel_sales'].fillna('noinfo',inplace=True) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "date_modif_prod: date of last modification of the product. Nearly 1% missing data in this variable may indicate that the product is not modified or the data is actually missing. \n",
    "date_end, date_renewal also has missing data.\n",
    "Can replace the missing values with the most common value. Cannot use median as this is a date data type. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "customer_data['date_modif_prod'].fillna(customer_data['date_modif_prod'].value_counts().index[0], inplace=True)\n",
    "customer_data['date_end'].fillna(customer_data['date_end'].value_counts().index[0], inplace=True)\n",
    "customer_data['date_renewal'].fillna(customer_data['date_renewal'].value_counts().index[0], inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "forecast_discount_energy, forecast_price_energy_p1, forecast_price_energy_p2 and forecast_price_pow_p1 variables have 1% missing data. Since all these variable are regrding forecasted values, we could first explore if missing values of each variable belong to same row (instance). "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "forecast_discount_energy    126\n",
       "forecast_price_energy_p1    126\n",
       "forecast_price_energy_p2    126\n",
       "forecast_price_pow_p1       126\n",
       "dtype: int64"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bool_null = pd.isnull(customer_data[['forecast_discount_energy', 'forecast_price_energy_p1', 'forecast_price_energy_p2','forecast_price_pow_p1']])\n",
    "bool_null.sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "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>forecast_discount_energy</th>\n",
       "      <th>forecast_price_energy_p1</th>\n",
       "      <th>forecast_price_energy_p2</th>\n",
       "      <th>forecast_price_pow_p1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>83</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>512</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>598</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>622</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15561</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15653</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15706</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15719</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15822</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>126 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       forecast_discount_energy  forecast_price_energy_p1  \\\n",
       "26                         True                      True   \n",
       "83                         True                      True   \n",
       "512                        True                      True   \n",
       "598                        True                      True   \n",
       "622                        True                      True   \n",
       "...                         ...                       ...   \n",
       "15561                      True                      True   \n",
       "15653                      True                      True   \n",
       "15706                      True                      True   \n",
       "15719                      True                      True   \n",
       "15822                      True                      True   \n",
       "\n",
       "       forecast_price_energy_p2  forecast_price_pow_p1  \n",
       "26                         True                   True  \n",
       "83                         True                   True  \n",
       "512                        True                   True  \n",
       "598                        True                   True  \n",
       "622                        True                   True  \n",
       "...                         ...                    ...  \n",
       "15561                      True                   True  \n",
       "15653                      True                   True  \n",
       "15706                      True                   True  \n",
       "15719                      True                   True  \n",
       "15822                      True                   True  \n",
       "\n",
       "[126 rows x 4 columns]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bool_null.loc[bool_null['forecast_discount_energy'] == True]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The rows (instances) can be removed from the dataset as missing data of each of these variable belong to the same row. Corresponding instances have to be removed from the other two datasets also.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "origin_up: code of the electricity campaign the customer first subscribed to. 1% data is missing. origin_up is a categorical variable. Missing data can be imputed by 'missingcampaign'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "customer_data['origin_up'].fillna('missingcampaign',inplace=True) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h5> Exploring price history data (pricing_data) </h5>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>price_date</th>\n",
       "      <th>price_p1_var</th>\n",
       "      <th>price_p2_var</th>\n",
       "      <th>price_p3_var</th>\n",
       "      <th>price_p1_fix</th>\n",
       "      <th>price_p2_fix</th>\n",
       "      <th>price_p3_fix</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>038af19179925da21a25619c5a24b745</td>\n",
       "      <td>2015-01-01</td>\n",
       "      <td>0.151367</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>44.266931</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>038af19179925da21a25619c5a24b745</td>\n",
       "      <td>2015-02-01</td>\n",
       "      <td>0.151367</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>44.266931</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>038af19179925da21a25619c5a24b745</td>\n",
       "      <td>2015-03-01</td>\n",
       "      <td>0.151367</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>44.266931</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>038af19179925da21a25619c5a24b745</td>\n",
       "      <td>2015-04-01</td>\n",
       "      <td>0.149626</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>44.266931</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>038af19179925da21a25619c5a24b745</td>\n",
       "      <td>2015-05-01</td>\n",
       "      <td>0.149626</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>44.266931</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 id  price_date  price_p1_var  price_p2_var  \\\n",
       "0  038af19179925da21a25619c5a24b745  2015-01-01      0.151367           0.0   \n",
       "1  038af19179925da21a25619c5a24b745  2015-02-01      0.151367           0.0   \n",
       "2  038af19179925da21a25619c5a24b745  2015-03-01      0.151367           0.0   \n",
       "3  038af19179925da21a25619c5a24b745  2015-04-01      0.149626           0.0   \n",
       "4  038af19179925da21a25619c5a24b745  2015-05-01      0.149626           0.0   \n",
       "\n",
       "   price_p3_var  price_p1_fix  price_p2_fix  price_p3_fix  \n",
       "0           0.0     44.266931           0.0           0.0  \n",
       "1           0.0     44.266931           0.0           0.0  \n",
       "2           0.0     44.266931           0.0           0.0  \n",
       "3           0.0     44.266931           0.0           0.0  \n",
       "4           0.0     44.266931           0.0           0.0  "
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pricing_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "id               object\n",
       "price_date       object\n",
       "price_p1_var    float64\n",
       "price_p2_var    float64\n",
       "price_p3_var    float64\n",
       "price_p1_fix    float64\n",
       "price_p2_fix    float64\n",
       "price_p3_fix    float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pricing_data.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Convert dates to datetime datatype\n",
    "pricing_data['price_date'] = pd.to_datetime(pricing_data['price_date'], format='%Y-%m-%d')"
   ]
  },
  {
   "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>price_p1_var</th>\n",
       "      <th>price_p2_var</th>\n",
       "      <th>price_p3_var</th>\n",
       "      <th>price_p1_fix</th>\n",
       "      <th>price_p2_fix</th>\n",
       "      <th>price_p3_fix</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>191643.000000</td>\n",
       "      <td>191643.000000</td>\n",
       "      <td>191643.000000</td>\n",
       "      <td>191643.000000</td>\n",
       "      <td>191643.000000</td>\n",
       "      <td>191643.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.140991</td>\n",
       "      <td>0.054412</td>\n",
       "      <td>0.030712</td>\n",
       "      <td>43.325546</td>\n",
       "      <td>10.698201</td>\n",
       "      <td>6.455436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.025117</td>\n",
       "      <td>0.050033</td>\n",
       "      <td>0.036335</td>\n",
       "      <td>5.437952</td>\n",
       "      <td>12.856046</td>\n",
       "      <td>7.782279</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.177779</td>\n",
       "      <td>-0.097752</td>\n",
       "      <td>-0.065172</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.125976</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>40.728885</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.146033</td>\n",
       "      <td>0.085483</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>44.266930</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.151635</td>\n",
       "      <td>0.101780</td>\n",
       "      <td>0.072558</td>\n",
       "      <td>44.444710</td>\n",
       "      <td>24.339581</td>\n",
       "      <td>16.226389</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>0.280700</td>\n",
       "      <td>0.229788</td>\n",
       "      <td>0.114102</td>\n",
       "      <td>59.444710</td>\n",
       "      <td>36.490692</td>\n",
       "      <td>17.458221</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        price_p1_var   price_p2_var   price_p3_var   price_p1_fix  \\\n",
       "count  191643.000000  191643.000000  191643.000000  191643.000000   \n",
       "mean        0.140991       0.054412       0.030712      43.325546   \n",
       "std         0.025117       0.050033       0.036335       5.437952   \n",
       "min         0.000000       0.000000       0.000000      -0.177779   \n",
       "25%         0.125976       0.000000       0.000000      40.728885   \n",
       "50%         0.146033       0.085483       0.000000      44.266930   \n",
       "75%         0.151635       0.101780       0.072558      44.444710   \n",
       "max         0.280700       0.229788       0.114102      59.444710   \n",
       "\n",
       "        price_p2_fix   price_p3_fix  \n",
       "count  191643.000000  191643.000000  \n",
       "mean       10.698201       6.455436  \n",
       "std        12.856046       7.782279  \n",
       "min        -0.097752      -0.065172  \n",
       "25%         0.000000       0.000000  \n",
       "50%         0.000000       0.000000  \n",
       "75%        24.339581      16.226389  \n",
       "max        36.490692      17.458221  "
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pricing_data.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There doesn't seem to be any missing data. Also, there doesn't seem to be outliers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "918f4508f7ed91cbac79a8905f61eca6    12\n",
       "73373ad47647846d301020fa4d0e8727    12\n",
       "a0f7d53eb0b442e512da9345491b70e5    12\n",
       "c1e657b3fb2d3a0ea1b6e0a12674a56b    12\n",
       "9296a3ca343aca9bf4bc6a971a8c7f76    12\n",
       "                                    ..\n",
       "3e459d61dc831e29f8a9a9a59f95efd2     8\n",
       "223a98d3832ece78cbf279a194868b54     8\n",
       "bf89f2d8c1b133a134fd93603cb4c947     7\n",
       "15b36e47cf04bf151e3f4438d12672e5     7\n",
       "c5dcd5c506e565aaabffa29bc1ec0a37     7\n",
       "Name: id, Length: 16096, dtype: int64"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pricing_data['id'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Length of the price history seem to be different for customers ranging between 12 to 7"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Missing data Percentage (%)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>price_date</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>price_p1_var</th>\n",
       "      <td>0.704138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>price_p2_var</th>\n",
       "      <td>0.704138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>price_p3_var</th>\n",
       "      <td>0.704138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>price_p1_fix</th>\n",
       "      <td>0.704138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>price_p2_fix</th>\n",
       "      <td>0.704138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>price_p3_fix</th>\n",
       "      <td>0.704138</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              Missing data Percentage (%)\n",
       "id                               0.000000\n",
       "price_date                       0.000000\n",
       "price_p1_var                     0.704138\n",
       "price_p2_var                     0.704138\n",
       "price_p3_var                     0.704138\n",
       "price_p1_fix                     0.704138\n",
       "price_p2_fix                     0.704138\n",
       "price_p3_fix                     0.704138"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame ({ \"Missing data Percentage (%)\" : pricing_data.isnull().sum()/len(pricing_data.index)*100})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This dataset has smaller percentages of missing data in all variables.\n",
    "Fill missing values with median:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "pricing_data['price_p1_var'].fillna(pricing_data['price_p1_var'].median(), inplace=True)\n",
    "pricing_data['price_p2_var'].fillna(pricing_data['price_p2_var'].median(), inplace=True)\n",
    "pricing_data['price_p3_var'].fillna(pricing_data['price_p3_var'].median(), inplace=True)\n",
    "pricing_data['price_p1_fix'].fillna(pricing_data['price_p1_fix'].median(), inplace=True)\n",
    "pricing_data['price_p2_fix'].fillna(pricing_data['price_p2_fix'].median(), inplace=True)\n",
    "pricing_data['price_p3_fix'].fillna(pricing_data['price_p3_fix'].median(), inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It was earlier observed in the basic statistics of pricing_data that the minimum value for some price related variables are negative values. So exploring it further:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No of rows with negative prices in price_p1_fix:  10\n",
      "No of rows with negative prices in price_p2_fix:  9\n",
      "No of rows with negative prices in price_p3_fix:  9\n"
     ]
    }
   ],
   "source": [
    "neg_price = pricing_data.apply(lambda x : True if x['price_p1_fix'] <0 else False, axis = 1) \n",
    "# count 'True' in neg_price series\n",
    "num_rows = len(neg_price[neg_price == True].index) \n",
    "print('No of rows with negative prices in price_p1_fix: ', num_rows ) \n",
    "\n",
    "neg_price = pricing_data.apply(lambda x : True if x['price_p2_fix'] <0 else False, axis = 1) \n",
    "# count 'True' in neg_price series\n",
    "num_rows = len(neg_price[neg_price == True].index) \n",
    "print('No of rows with negative prices in price_p2_fix: ', num_rows ) \n",
    "\n",
    "neg_price = pricing_data.apply(lambda x : True if x['price_p3_fix'] <0 else False, axis = 1) \n",
    "# count 'True' in neg_price series\n",
    "num_rows = len(neg_price[neg_price == True].index) \n",
    "print('No of rows with negative prices in price_p3_fix: ', num_rows ) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Replacing the negative prices with median of corresponding variable values\n",
    "pricing_data.loc[pricing_data['price_p1_fix']<0,'price_p1_fix'] = pricing_data['price_p1_fix'].median()\n",
    "pricing_data.loc[pricing_data['price_p2_fix']<0,'price_p2_fix'] = pricing_data['price_p2_fix'].median()\n",
    "pricing_data.loc[pricing_data['price_p3_fix']<0,'price_p3_fix'] = pricing_data['price_p3_fix'].median()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "#serializing the pre-processed dataframes to use later\n",
    "pd.to_pickle(customer_data,'Data\\ProcessedData\\customer_data.pkl')\n",
    "pd.to_pickle(pricing_data,'Data\\ProcessedData\\pricing_data.pkl')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}