Pareto_NBD.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 254,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#Load libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import scipy as sp\n",
    "from datetime import datetime"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 255,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 258,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(146874, 16)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>exchange_rate_date</th>\n",
       "      <th>invoice</th>\n",
       "      <th>customer</th>\n",
       "      <th>id</th>\n",
       "      <th>state</th>\n",
       "      <th>order</th>\n",
       "      <th>createdAt</th>\n",
       "      <th>location</th>\n",
       "      <th>segmentation</th>\n",
       "      <th>NetTotalAfterDiscount</th>\n",
       "      <th>NetTotalBeforeDiscount</th>\n",
       "      <th>GrossTotalAfterDiscount</th>\n",
       "      <th>GrossTotalBeforeDiscount</th>\n",
       "      <th>currency_code</th>\n",
       "      <th>exchange_rate_value</th>\n",
       "      <th>isvalid</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2014-10-22</td>\n",
       "      <td>NaN</td>\n",
       "      <td>GB-CUS-Y8922200</td>\n",
       "      <td>54478738d2cde8320d9b58d2</td>\n",
       "      <td>canceled</td>\n",
       "      <td>GB-A-69332</td>\n",
       "      <td>2014-10-22 12:30:16</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>GBP</td>\n",
       "      <td>0.788134</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2014-10-22</td>\n",
       "      <td>NaN</td>\n",
       "      <td>GB-CUS-K3854440</td>\n",
       "      <td>54478fa5d4cde82847f1345e</td>\n",
       "      <td>canceled</td>\n",
       "      <td>GB-A-74327</td>\n",
       "      <td>2014-10-22 13:06:13</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>DCLA</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>GBP</td>\n",
       "      <td>0.788134</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  exchange_rate_date invoice         customer                        id  \\\n",
       "0         2014-10-22     NaN  GB-CUS-Y8922200  54478738d2cde8320d9b58d2   \n",
       "1         2014-10-22     NaN  GB-CUS-K3854440  54478fa5d4cde82847f1345e   \n",
       "\n",
       "      state       order            createdAt   location segmentation  \\\n",
       "0  canceled  GB-A-69332  2014-10-22 12:30:16  gb_london          NaN   \n",
       "1  canceled  GB-A-74327  2014-10-22 13:06:13  gb_london         DCLA   \n",
       "\n",
       "   NetTotalAfterDiscount  NetTotalBeforeDiscount  GrossTotalAfterDiscount  \\\n",
       "0                    NaN                     NaN                      NaN   \n",
       "1                    NaN                     NaN                      NaN   \n",
       "\n",
       "   GrossTotalBeforeDiscount currency_code  exchange_rate_value  isvalid  \n",
       "0                       NaN           GBP             0.788134        0  \n",
       "1                       NaN           GBP             0.788134        0  "
      ]
     },
     "execution_count": 258,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Load data\n",
    "data = pd.read_csv(\"orders_data.csv\")\n",
    "print(data.shape)\n",
    "data.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 259,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#Make a subset\n",
    "data_s = data[[\"customer\",\"state\",\"order\",\"createdAt\",\"location\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "customer             object\n",
      "state                object\n",
      "order                object\n",
      "createdAt            object\n",
      "location             object\n",
      "date         datetime64[ns]\n",
      "dtype: object\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>customer</th>\n",
       "      <th>state</th>\n",
       "      <th>order</th>\n",
       "      <th>createdAt</th>\n",
       "      <th>location</th>\n",
       "      <th>date</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GB-CUS-Y8922200</td>\n",
       "      <td>canceled</td>\n",
       "      <td>GB-A-69332</td>\n",
       "      <td>2014-10-22 12:30:16</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>2014-10-22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>GB-CUS-K3854440</td>\n",
       "      <td>canceled</td>\n",
       "      <td>GB-A-74327</td>\n",
       "      <td>2014-10-22 13:06:13</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>2014-10-22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>GB-CUS-Y8922200</td>\n",
       "      <td>canceled</td>\n",
       "      <td>GB-R-37069</td>\n",
       "      <td>2014-10-22 12:34:35</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>2014-10-22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>GB-CUS-C3589820</td>\n",
       "      <td>canceled</td>\n",
       "      <td>GB-G-10354</td>\n",
       "      <td>2014-10-22 20:04:20</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>2014-10-22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>GB-CUS-P6345195</td>\n",
       "      <td>completed</td>\n",
       "      <td>GB-X-54771</td>\n",
       "      <td>2014-10-22 12:24:29</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>2014-10-22</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          customer      state       order            createdAt   location  \\\n",
       "0  GB-CUS-Y8922200   canceled  GB-A-69332  2014-10-22 12:30:16  gb_london   \n",
       "1  GB-CUS-K3854440   canceled  GB-A-74327  2014-10-22 13:06:13  gb_london   \n",
       "2  GB-CUS-Y8922200   canceled  GB-R-37069  2014-10-22 12:34:35  gb_london   \n",
       "3  GB-CUS-C3589820   canceled  GB-G-10354  2014-10-22 20:04:20  gb_london   \n",
       "4  GB-CUS-P6345195  completed  GB-X-54771  2014-10-22 12:24:29  gb_london   \n",
       "\n",
       "        date  \n",
       "0 2014-10-22  \n",
       "1 2014-10-22  \n",
       "2 2014-10-22  \n",
       "3 2014-10-22  \n",
       "4 2014-10-22  "
      ]
     },
     "execution_count": 260,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Add the data, just in case\n",
    "data_s.is_copy = False #turn off the check for that object\n",
    "#data_s[\"date\"] = pd.to_datetime(data_s[\"createdAt\"]) #that form would keep the timestamp\n",
    "data_s[\"date\"] = pd.to_datetime(data_s[\"createdAt\"]).dt.date #that parses the date part \n",
    "data_s['date'] = pd.to_datetime(data_s['date'])\n",
    "print(data_s.dtypes)\n",
    "data_s.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['canceled' 'completed' 'payment_error' 'processing'\n",
      " 'payment_settlement_cycle' 'reserved']\n",
      "Total number of valid orders:  123970\n",
      "Unique customers:  37310\n",
      "(146874, 7)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>customer</th>\n",
       "      <th>state</th>\n",
       "      <th>order</th>\n",
       "      <th>createdAt</th>\n",
       "      <th>location</th>\n",
       "      <th>date</th>\n",
       "      <th>isvalid</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GB-CUS-Y8922200</td>\n",
       "      <td>canceled</td>\n",
       "      <td>GB-A-69332</td>\n",
       "      <td>2014-10-22 12:30:16</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>2014-10-22</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>GB-CUS-K3854440</td>\n",
       "      <td>canceled</td>\n",
       "      <td>GB-A-74327</td>\n",
       "      <td>2014-10-22 13:06:13</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>2014-10-22</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>GB-CUS-Y8922200</td>\n",
       "      <td>canceled</td>\n",
       "      <td>GB-R-37069</td>\n",
       "      <td>2014-10-22 12:34:35</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>2014-10-22</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>GB-CUS-C3589820</td>\n",
       "      <td>canceled</td>\n",
       "      <td>GB-G-10354</td>\n",
       "      <td>2014-10-22 20:04:20</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>2014-10-22</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>GB-CUS-P6345195</td>\n",
       "      <td>completed</td>\n",
       "      <td>GB-X-54771</td>\n",
       "      <td>2014-10-22 12:24:29</td>\n",
       "      <td>gb_london</td>\n",
       "      <td>2014-10-22</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          customer      state       order            createdAt   location  \\\n",
       "0  GB-CUS-Y8922200   canceled  GB-A-69332  2014-10-22 12:30:16  gb_london   \n",
       "1  GB-CUS-K3854440   canceled  GB-A-74327  2014-10-22 13:06:13  gb_london   \n",
       "2  GB-CUS-Y8922200   canceled  GB-R-37069  2014-10-22 12:34:35  gb_london   \n",
       "3  GB-CUS-C3589820   canceled  GB-G-10354  2014-10-22 20:04:20  gb_london   \n",
       "4  GB-CUS-P6345195  completed  GB-X-54771  2014-10-22 12:24:29  gb_london   \n",
       "\n",
       "        date  isvalid  \n",
       "0 2014-10-22        0  \n",
       "1 2014-10-22        0  \n",
       "2 2014-10-22        0  \n",
       "3 2014-10-22        0  \n",
       "4 2014-10-22        1  "
      ]
     },
     "execution_count": 261,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Check the set of unique states\n",
    "print(data_s['state'].unique())\n",
    "\n",
    "#Add is_valid column\n",
    "data_s[\"isvalid\"] = 1\n",
    "data_s.ix[data_s['state'].isin(['new','canceled','reserved']),['isvalid']] = 0\n",
    "\n",
    "#Check total # of valid orders\n",
    "print(\"Total number of valid orders: \", sum(data_s[\"isvalid\"]))\n",
    "print(\"Unique customers: \", len(data_s['customer'].unique()))\n",
    "print(data_s.shape)\n",
    "data_s.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Preparation part"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#Aggregate by customer\n",
    "g = data_s[data_s['isvalid']>0].groupby('customer').apply(lambda x: x['isvalid'].sum()).reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 263,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "123970\n",
      "(33722, 2)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>customer</th>\n",
       "      <th>validOrders</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>CUS-A0013083</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>CUS-A0016193</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>CUS-A0022539</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>CUS-A0028640</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>CUS-A0032874</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       customer  validOrders\n",
       "0  CUS-A0013083            3\n",
       "1  CUS-A0016193            1\n",
       "2  CUS-A0022539            1\n",
       "3  CUS-A0028640            1\n",
       "4  CUS-A0032874            8"
      ]
     },
     "execution_count": 263,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "g.columns = ['customer','validOrders']\n",
    "print(g[\"validOrders\"].sum())\n",
    "print(g.shape)\n",
    "g.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Make a transform with first and last date per customer\n",
    "#p = data_s[data_s['isvalid']>0].groupby('customer')['date'].agg({\"first\": lambda x: x.iloc[0],\"last\": lambda x: x.iloc[-1]}).reset_index()\n",
    "p = data_s[data_s['isvalid']>0].groupby('customer')['date'].agg({\"first\": lambda x: x.min(),\"last\": lambda x: x.max()}).reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 266,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(33722, 3)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>customer</th>\n",
       "      <th>first</th>\n",
       "      <th>last</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>CUS-A0013083</td>\n",
       "      <td>2015-10-12</td>\n",
       "      <td>2016-09-27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>CUS-A0016193</td>\n",
       "      <td>2016-12-12</td>\n",
       "      <td>2016-12-12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>CUS-A0022539</td>\n",
       "      <td>2016-01-06</td>\n",
       "      <td>2016-01-06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>CUS-A0028640</td>\n",
       "      <td>2016-09-09</td>\n",
       "      <td>2016-09-09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>CUS-A0032874</td>\n",
       "      <td>2016-03-31</td>\n",
       "      <td>2016-05-14</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       customer      first       last\n",
       "0  CUS-A0013083 2015-10-12 2016-09-27\n",
       "1  CUS-A0016193 2016-12-12 2016-12-12\n",
       "2  CUS-A0022539 2016-01-06 2016-01-06\n",
       "3  CUS-A0028640 2016-09-09 2016-09-09\n",
       "4  CUS-A0032874 2016-03-31 2016-05-14"
      ]
     },
     "execution_count": 266,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check shape \n",
    "print(p.shape)\n",
    "p.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 267,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(33722, 4)\n",
      "123970\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>customer</th>\n",
       "      <th>validOrders</th>\n",
       "      <th>first</th>\n",
       "      <th>last</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>CUS-A0013083</td>\n",
       "      <td>3</td>\n",
       "      <td>2015-10-12</td>\n",
       "      <td>2016-09-27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>CUS-A0016193</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-12-12</td>\n",
       "      <td>2016-12-12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>CUS-A0022539</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-06</td>\n",
       "      <td>2016-01-06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>CUS-A0028640</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-09-09</td>\n",
       "      <td>2016-09-09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>CUS-A0032874</td>\n",
       "      <td>8</td>\n",
       "      <td>2016-03-31</td>\n",
       "      <td>2016-05-14</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       customer  validOrders      first       last\n",
       "0  CUS-A0013083            3 2015-10-12 2016-09-27\n",
       "1  CUS-A0016193            1 2016-12-12 2016-12-12\n",
       "2  CUS-A0022539            1 2016-01-06 2016-01-06\n",
       "3  CUS-A0028640            1 2016-09-09 2016-09-09\n",
       "4  CUS-A0032874            8 2016-03-31 2016-05-14"
      ]
     },
     "execution_count": 267,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Merge the datasets\n",
    "merged_df = pd.merge(left=g,right=p,how='inner',left_on='customer',right_on='customer',copy=True)\n",
    "print(merged_df.shape)\n",
    "print(merged_df['validOrders'].sum())\n",
    "merged_df.head()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 268,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>customer</th>\n",
       "      <th>validOrders</th>\n",
       "      <th>first</th>\n",
       "      <th>last</th>\n",
       "      <th>frequency</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>CUS-A0013083</td>\n",
       "      <td>3</td>\n",
       "      <td>2015-10-12</td>\n",
       "      <td>2016-09-27</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>CUS-A0016193</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-12-12</td>\n",
       "      <td>2016-12-12</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>CUS-A0022539</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-06</td>\n",
       "      <td>2016-01-06</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>CUS-A0028640</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-09-09</td>\n",
       "      <td>2016-09-09</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>CUS-A0032874</td>\n",
       "      <td>8</td>\n",
       "      <td>2016-03-31</td>\n",
       "      <td>2016-05-14</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       customer  validOrders      first       last  frequency\n",
       "0  CUS-A0013083            3 2015-10-12 2016-09-27          2\n",
       "1  CUS-A0016193            1 2016-12-12 2016-12-12          0\n",
       "2  CUS-A0022539            1 2016-01-06 2016-01-06          0\n",
       "3  CUS-A0028640            1 2016-09-09 2016-09-09          0\n",
       "4  CUS-A0032874            8 2016-03-31 2016-05-14          7"
      ]
     },
     "execution_count": 268,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Get frequency\n",
    "merged_df['frequency'] = merged_df['validOrders'] - 1\n",
    "merged_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 269,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "customer               object\n",
      "validOrders             int64\n",
      "first          datetime64[ns]\n",
      "last           datetime64[ns]\n",
      "frequency               int64\n",
      "recency               float64\n",
      "dtype: object\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>customer</th>\n",
       "      <th>validOrders</th>\n",
       "      <th>first</th>\n",
       "      <th>last</th>\n",
       "      <th>frequency</th>\n",
       "      <th>recency</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [customer, validOrders, first, last, frequency, recency]\n",
       "Index: []"
      ]
     },
     "execution_count": 269,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Get recency: duration btw first and last purchase\n",
    "merged_df['recency'] = (merged_df['last'] - merged_df['first']).astype('timedelta64[W]') #D for days, W for weeks etc\n",
    "print(merged_df.dtypes)\n",
    "merged_df.head()\n",
    "\n",
    "#Check for negative values\n",
    "merged_df[merged_df['recency'] <0] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 270,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>customer</th>\n",
       "      <th>validOrders</th>\n",
       "      <th>first</th>\n",
       "      <th>last</th>\n",
       "      <th>frequency</th>\n",
       "      <th>recency</th>\n",
       "      <th>T</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>CUS-A0013083</td>\n",
       "      <td>3</td>\n",
       "      <td>2015-10-12</td>\n",
       "      <td>2016-09-27</td>\n",
       "      <td>2</td>\n",
       "      <td>50.0</td>\n",
       "      <td>72.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>CUS-A0016193</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-12-12</td>\n",
       "      <td>2016-12-12</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>11.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>CUS-A0022539</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-06</td>\n",
       "      <td>2016-01-06</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>60.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>CUS-A0028640</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-09-09</td>\n",
       "      <td>2016-09-09</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>24.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>CUS-A0032874</td>\n",
       "      <td>8</td>\n",
       "      <td>2016-03-31</td>\n",
       "      <td>2016-05-14</td>\n",
       "      <td>7</td>\n",
       "      <td>6.0</td>\n",
       "      <td>48.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       customer  validOrders      first       last  frequency  recency     T\n",
       "0  CUS-A0013083            3 2015-10-12 2016-09-27          2     50.0  72.0\n",
       "1  CUS-A0016193            1 2016-12-12 2016-12-12          0      0.0  11.0\n",
       "2  CUS-A0022539            1 2016-01-06 2016-01-06          0      0.0  60.0\n",
       "3  CUS-A0028640            1 2016-09-09 2016-09-09          0      0.0  24.0\n",
       "4  CUS-A0032874            8 2016-03-31 2016-05-14          7      6.0  48.0"
      ]
     },
     "execution_count": 270,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged_df[\"T\"] = (pd.datetime.now().date() - merged_df['first']).astype('timedelta64[W]')\n",
    "merged_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 271,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(33722, 3)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>frequency</th>\n",
       "      <th>recency</th>\n",
       "      <th>T</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>customer</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>CUS-A0013083</th>\n",
       "      <td>2</td>\n",
       "      <td>50.0</td>\n",
       "      <td>72.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-A0016193</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>11.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-A0022539</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>60.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-A0028640</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>24.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-A0032874</th>\n",
       "      <td>7</td>\n",
       "      <td>6.0</td>\n",
       "      <td>48.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              frequency  recency     T\n",
       "customer                              \n",
       "CUS-A0013083          2     50.0  72.0\n",
       "CUS-A0016193          0      0.0  11.0\n",
       "CUS-A0022539          0      0.0  60.0\n",
       "CUS-A0028640          0      0.0  24.0\n",
       "CUS-A0032874          7      6.0  48.0"
      ]
     },
     "execution_count": 271,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m = merged_df[['customer','frequency','recency','T']].set_index('customer')\n",
    "print(m.shape)\n",
    "m.head()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model part\n",
    "\n",
    "<p> Fitting the BG/NBD model </p>\n",
    "<p> </p>\n",
    "<p> Some background info: model framework was laid by Pareto/NBD model (1987) which has 2 parts: time to dropout is modeled using Pareto (exponential gamma mixture) dist, while customers being still alive their repeat purchase behaviour is modeled using negative binomial (Poisson gamma) distribution. Problem is with parameter fitting </p> \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from lifetimes import BetaGeoFitter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<lifetimes.BetaGeoFitter: fitted with 33722 subjects, a: 0.60, alpha: 2.20, b: 2.70, r: 0.22>\n"
     ]
    }
   ],
   "source": [
    "# similar API to scikit-learn and lifelines.\n",
    "bgf = BetaGeoFitter(penalizer_coef=0.0)\n",
    "bgf.fit(m['frequency'], m['recency'], m['T'])\n",
    "print(bgf)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 274,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12db57630>"
      ]
     },
     "execution_count": 274,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWcAAAElCAYAAAA4KCPqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x108b1c0f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from lifetimes.plotting import plot_frequency_recency_matrix\n",
    "plot_frequency_recency_matrix(bgf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x123255a20>"
      ]
     },
     "execution_count": 275,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUIAAAElCAYAAACRXOt+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nD1p8AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1055292e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from lifetimes.plotting import plot_probability_alive_matrix\n",
    "\n",
    "plot_probability_alive_matrix(bgf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 276,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>frequency</th>\n",
       "      <th>recency</th>\n",
       "      <th>T</th>\n",
       "      <th>predicted_purchases</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>customer</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>CUS-D3250964</th>\n",
       "      <td>50</td>\n",
       "      <td>52.0</td>\n",
       "      <td>52.0</td>\n",
       "      <td>3.589551</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-CN-04423160</th>\n",
       "      <td>51</td>\n",
       "      <td>53.0</td>\n",
       "      <td>53.0</td>\n",
       "      <td>3.596733</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-XS-92125017</th>\n",
       "      <td>53</td>\n",
       "      <td>55.0</td>\n",
       "      <td>55.0</td>\n",
       "      <td>3.610367</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-SF-00895675</th>\n",
       "      <td>42</td>\n",
       "      <td>42.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>3.676830</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-VN-03185864</th>\n",
       "      <td>55</td>\n",
       "      <td>56.0</td>\n",
       "      <td>56.0</td>\n",
       "      <td>3.684190</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-M6004676</th>\n",
       "      <td>71</td>\n",
       "      <td>72.0</td>\n",
       "      <td>72.0</td>\n",
       "      <td>3.750390</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-KR-76690240</th>\n",
       "      <td>38</td>\n",
       "      <td>32.0</td>\n",
       "      <td>33.0</td>\n",
       "      <td>4.029404</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-DS-33226215</th>\n",
       "      <td>67</td>\n",
       "      <td>58.0</td>\n",
       "      <td>59.0</td>\n",
       "      <td>4.202991</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-H2911406</th>\n",
       "      <td>86</td>\n",
       "      <td>76.0</td>\n",
       "      <td>76.0</td>\n",
       "      <td>4.317024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-P1335085</th>\n",
       "      <td>56</td>\n",
       "      <td>45.0</td>\n",
       "      <td>45.0</td>\n",
       "      <td>4.605682</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-Y9163636</th>\n",
       "      <td>46</td>\n",
       "      <td>34.0</td>\n",
       "      <td>35.0</td>\n",
       "      <td>4.622101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-TH-24036964</th>\n",
       "      <td>51</td>\n",
       "      <td>36.0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>4.874476</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-DX-87320253</th>\n",
       "      <td>102</td>\n",
       "      <td>67.0</td>\n",
       "      <td>68.0</td>\n",
       "      <td>5.591283</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-ZV-35265289</th>\n",
       "      <td>17</td>\n",
       "      <td>7.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>6.552997</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-FA-71022404</th>\n",
       "      <td>205</td>\n",
       "      <td>69.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>10.652125</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   frequency  recency     T  predicted_purchases\n",
       "customer                                                        \n",
       "CUS-D3250964              50     52.0  52.0             3.589551\n",
       "CUS-W-CN-04423160         51     53.0  53.0             3.596733\n",
       "CUS-W-XS-92125017         53     55.0  55.0             3.610367\n",
       "CUS-W-SF-00895675         42     42.0  42.0             3.676830\n",
       "CUS-W-VN-03185864         55     56.0  56.0             3.684190\n",
       "CUS-M6004676              71     72.0  72.0             3.750390\n",
       "CUS-W-KR-76690240         38     32.0  33.0             4.029404\n",
       "CUS-W-DS-33226215         67     58.0  59.0             4.202991\n",
       "CUS-H2911406              86     76.0  76.0             4.317024\n",
       "CUS-P1335085              56     45.0  45.0             4.605682\n",
       "CUS-Y9163636              46     34.0  35.0             4.622101\n",
       "CUS-W-TH-24036964         51     36.0  37.0             4.874476\n",
       "CUS-W-DX-87320253        102     67.0  68.0             5.591283\n",
       "CUS-W-ZV-35265289         17      7.0   7.0             6.552997\n",
       "CUS-W-FA-71022404        205     69.0  70.0            10.652125"
      ]
     },
     "execution_count": 276,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "t = 4\n",
    "m['predicted_purchases'] = bgf.conditional_expected_number_of_purchases_up_to_time(t, m['frequency'], m['recency'], m['T'])\n",
    "m.sort_values(by='predicted_purchases').tail(15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 277,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4974.4719218408291"
      ]
     },
     "execution_count": 277,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m['predicted_purchases'].sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 278,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>frequency</th>\n",
       "      <th>recency</th>\n",
       "      <th>T</th>\n",
       "      <th>predicted_purchases</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>customer</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>CUS-A0013083</th>\n",
       "      <td>2</td>\n",
       "      <td>50.0</td>\n",
       "      <td>72.0</td>\n",
       "      <td>0.087832</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-A0016193</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0.065445</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-A0022539</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>60.0</td>\n",
       "      <td>0.014235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-A0028640</th>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>0.033479</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-A0032874</th>\n",
       "      <td>7</td>\n",
       "      <td>6.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>0.000017</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              frequency  recency     T  predicted_purchases\n",
       "customer                                                   \n",
       "CUS-A0013083          2     50.0  72.0             0.087832\n",
       "CUS-A0016193          0      0.0  11.0             0.065445\n",
       "CUS-A0022539          0      0.0  60.0             0.014235\n",
       "CUS-A0028640          0      0.0  24.0             0.033479\n",
       "CUS-A0032874          7      6.0  48.0             0.000017"
      ]
     },
     "execution_count": 278,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m.reset_index()\n",
    "m.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Backtest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 279,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Timestamp('2016-12-31 00:00:00')"
      ]
     },
     "execution_count": 279,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Make a subset \n",
    "data_test = data_s[data_s['date'] < datetime(2017,1,1)]\n",
    "\n",
    "#Check the latest date\n",
    "data_test['date'].max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 287,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "g_test = data_test[data_test['isvalid']>0].groupby('customer').apply(lambda x: x['isvalid'].sum()).reset_index()\n",
    "g_test.columns = ['customer','validOrders']\n",
    "p_test = data_test[data_test['isvalid']>0].groupby('customer')['date'].agg({\"first\": lambda x: x.min(),\"last\": lambda x: x.max()}).reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(30725, 2)\n",
      "(30725, 3)\n",
      "(30725, 4)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>customer</th>\n",
       "      <th>validOrders</th>\n",
       "      <th>first</th>\n",
       "      <th>last</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>CUS-A0013083</td>\n",
       "      <td>3</td>\n",
       "      <td>2015-10-12</td>\n",
       "      <td>2016-09-27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>CUS-A0016193</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-12-12</td>\n",
       "      <td>2016-12-12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>CUS-A0022539</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-06</td>\n",
       "      <td>2016-01-06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>CUS-A0028640</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-09-09</td>\n",
       "      <td>2016-09-09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>CUS-A0032874</td>\n",
       "      <td>8</td>\n",
       "      <td>2016-03-31</td>\n",
       "      <td>2016-05-14</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       customer  validOrders      first       last\n",
       "0  CUS-A0013083            3 2015-10-12 2016-09-27\n",
       "1  CUS-A0016193            1 2016-12-12 2016-12-12\n",
       "2  CUS-A0022539            1 2016-01-06 2016-01-06\n",
       "3  CUS-A0028640            1 2016-09-09 2016-09-09\n",
       "4  CUS-A0032874            8 2016-03-31 2016-05-14"
      ]
     },
     "execution_count": 288,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(g_test.shape)\n",
    "print(p_test.shape)\n",
    "merged_test = pd.merge(left=g_test,right=p_test,how='inner',left_on='customer',right_on='customer',copy=True)\n",
    "print(merged_test.shape)\n",
    "merged_test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 297,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30725, 7)"
      ]
     },
     "execution_count": 297,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged_test[\"T\"] = (pd.datetime(2017,1,1) - merged_test['first']).astype('timedelta64[W]')\n",
    "merged_test['recency'] = (merged_test['last'] - merged_test['first']).astype('timedelta64[W]')\n",
    "merged_test['frequency'] = merged_test['validOrders'] - 1\n",
    "merged_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 298,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30725, 3)"
      ]
     },
     "execution_count": 298,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m_test = merged_test[['customer','frequency','recency','T']].set_index('customer')\n",
    "m_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 299,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<lifetimes.BetaGeoFitter: fitted with 30725 subjects, a: 0.62, alpha: 2.28, b: 2.85, r: 0.23>\n"
     ]
    }
   ],
   "source": [
    "bgf_test = BetaGeoFitter(penalizer_coef=0.0)\n",
    "bgf_test.fit(m_test['frequency'], m_test['recency'], m_test['T'])\n",
    "print(bgf_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 300,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>frequency</th>\n",
       "      <th>recency</th>\n",
       "      <th>T</th>\n",
       "      <th>predicted_purchases</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>customer</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>CUS-W-VN-03185864</th>\n",
       "      <td>46</td>\n",
       "      <td>47.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>3.482096</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-M6004676</th>\n",
       "      <td>62</td>\n",
       "      <td>62.0</td>\n",
       "      <td>63.0</td>\n",
       "      <td>3.653980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-V5982835</th>\n",
       "      <td>49</td>\n",
       "      <td>46.0</td>\n",
       "      <td>47.0</td>\n",
       "      <td>3.779015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-H2911406</th>\n",
       "      <td>74</td>\n",
       "      <td>66.0</td>\n",
       "      <td>68.0</td>\n",
       "      <td>3.886419</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-P1335085</th>\n",
       "      <td>40</td>\n",
       "      <td>36.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>4.023221</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-KR-76690240</th>\n",
       "      <td>29</td>\n",
       "      <td>24.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>4.185678</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-DS-33226215</th>\n",
       "      <td>57</td>\n",
       "      <td>50.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>4.239196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-C3489321</th>\n",
       "      <td>37</td>\n",
       "      <td>28.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>4.363398</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-F4899355</th>\n",
       "      <td>16</td>\n",
       "      <td>11.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>4.386858</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-TH-24036964</th>\n",
       "      <td>40</td>\n",
       "      <td>28.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>5.048698</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-YX-36703435</th>\n",
       "      <td>47</td>\n",
       "      <td>31.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>5.074367</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-Y9163636</th>\n",
       "      <td>38</td>\n",
       "      <td>26.0</td>\n",
       "      <td>26.0</td>\n",
       "      <td>5.121100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-FA-71022404</th>\n",
       "      <td>183</td>\n",
       "      <td>59.0</td>\n",
       "      <td>61.0</td>\n",
       "      <td>5.174067</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-W-DX-87320253</th>\n",
       "      <td>92</td>\n",
       "      <td>57.0</td>\n",
       "      <td>59.0</td>\n",
       "      <td>5.181475</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CUS-P0788059</th>\n",
       "      <td>32</td>\n",
       "      <td>14.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>6.240237</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   frequency  recency     T  predicted_purchases\n",
       "customer                                                        \n",
       "CUS-W-VN-03185864         46     47.0  48.0             3.482096\n",
       "CUS-M6004676              62     62.0  63.0             3.653980\n",
       "CUS-V5982835              49     46.0  47.0             3.779015\n",
       "CUS-H2911406              74     66.0  68.0             3.886419\n",
       "CUS-P1335085              40     36.0  36.0             4.023221\n",
       "CUS-W-KR-76690240         29     24.0  24.0             4.185678\n",
       "CUS-W-DS-33226215         57     50.0  50.0             4.239196\n",
       "CUS-C3489321              37     28.0  29.0             4.363398\n",
       "CUS-F4899355              16     11.0  11.0             4.386858\n",
       "CUS-W-TH-24036964         40     28.0  28.0             5.048698\n",
       "CUS-W-YX-36703435         47     31.0  32.0             5.074367\n",
       "CUS-Y9163636              38     26.0  26.0             5.121100\n",
       "CUS-W-FA-71022404        183     59.0  61.0             5.174067\n",
       "CUS-W-DX-87320253         92     57.0  59.0             5.181475\n",
       "CUS-P0788059              32     14.0  15.0             6.240237"
      ]
     },
     "execution_count": 300,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "t = 4\n",
    "m_test['predicted_purchases'] = bgf_test.conditional_expected_number_of_purchases_up_to_time(t, m_test['frequency'], m_test['recency'], m_test['T'])\n",
    "m_test.sort_values(by='predicted_purchases').tail(15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 309,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0968169041793\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "4423.7100846204494"
      ]
     },
     "execution_count": 309,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print((3270+1253+329 - m_test['predicted_purchases'].sum())/m_test['predicted_purchases'].sum())\n",
    "m_test['predicted_purchases'].sum()"
   ]
  }
 ],
 "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": "3.5.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}