Retail Rocket Ecommerce.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/Users/meganyow/Documents/Interviews/Shopify'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "os.getcwd()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "cat_tree = pd.read_csv(\"ecommerce-dataset/category_tree.csv\")\n",
    "events = pd.read_csv(\"ecommerce-dataset/events.csv\")\n",
    "item_prop1 = pd.read_csv(\"ecommerce-dataset/item_properties_part1.csv\")\n",
    "item_prop2 = pd.read_csv(\"ecommerce-dataset/item_properties_part2.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "item_prop = pd.concat([item_prop1, item_prop2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>categoryid</th>\n",
       "      <th>parentid</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1016</td>\n",
       "      <td>213.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>809</td>\n",
       "      <td>169.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>570</td>\n",
       "      <td>9.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1691</td>\n",
       "      <td>885.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>536</td>\n",
       "      <td>1691.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   categoryid  parentid\n",
       "0        1016     213.0\n",
       "1         809     169.0\n",
       "2         570       9.0\n",
       "3        1691     885.0\n",
       "4         536    1691.0"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cat_tree.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "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>timestamp</th>\n",
       "      <th>visitorid</th>\n",
       "      <th>event</th>\n",
       "      <th>itemid</th>\n",
       "      <th>transactionid</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1433221332117</td>\n",
       "      <td>257597</td>\n",
       "      <td>view</td>\n",
       "      <td>355908</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1433224214164</td>\n",
       "      <td>992329</td>\n",
       "      <td>view</td>\n",
       "      <td>248676</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1433221999827</td>\n",
       "      <td>111016</td>\n",
       "      <td>view</td>\n",
       "      <td>318965</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1433221955914</td>\n",
       "      <td>483717</td>\n",
       "      <td>view</td>\n",
       "      <td>253185</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1433221337106</td>\n",
       "      <td>951259</td>\n",
       "      <td>view</td>\n",
       "      <td>367447</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       timestamp  visitorid event  itemid  transactionid\n",
       "0  1433221332117     257597  view  355908            NaN\n",
       "1  1433224214164     992329  view  248676            NaN\n",
       "2  1433221999827     111016  view  318965            NaN\n",
       "3  1433221955914     483717  view  253185            NaN\n",
       "4  1433221337106     951259  view  367447            NaN"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "events.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "item_prop['new_value'] = item_prop1.value.apply(lambda x: x[1:].split(\" \", 1)[0] if x.startswith('n') else np.nan)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Events Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "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>timestamp</th>\n",
       "      <th>visitorid</th>\n",
       "      <th>event</th>\n",
       "      <th>itemid</th>\n",
       "      <th>transactionid</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>event</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>addtocart</th>\n",
       "      <td>68959</td>\n",
       "      <td>37722</td>\n",
       "      <td>1</td>\n",
       "      <td>23903</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>transaction</th>\n",
       "      <td>20362</td>\n",
       "      <td>11719</td>\n",
       "      <td>1</td>\n",
       "      <td>12025</td>\n",
       "      <td>17672</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>view</th>\n",
       "      <td>2661325</td>\n",
       "      <td>1404179</td>\n",
       "      <td>1</td>\n",
       "      <td>234838</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             timestamp  visitorid  event  itemid  transactionid\n",
       "event                                                          \n",
       "addtocart        68959      37722      1   23903              0\n",
       "transaction      20362      11719      1   12025          17672\n",
       "view           2661325    1404179      1  234838              0"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "events.groupby(['event']).nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "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>timestamp</th>\n",
       "      <th>visitorid</th>\n",
       "      <th>itemid</th>\n",
       "      <th>transactionid</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>event</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>addtocart</th>\n",
       "      <td>69332</td>\n",
       "      <td>69332</td>\n",
       "      <td>69332</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>transaction</th>\n",
       "      <td>22457</td>\n",
       "      <td>22457</td>\n",
       "      <td>22457</td>\n",
       "      <td>22457</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>view</th>\n",
       "      <td>2664312</td>\n",
       "      <td>2664312</td>\n",
       "      <td>2664312</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             timestamp  visitorid   itemid  transactionid\n",
       "event                                                    \n",
       "addtocart        69332      69332    69332              0\n",
       "transaction      22457      22457    22457          22457\n",
       "view           2664312    2664312  2664312              0"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "events.groupby(['event']).count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "78.7% of transactions are single item so a market basket analysis not suitable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7869261254842588"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "17672 / 22457\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Visitor count on actions: (2756101,)\n",
      "Total unique visitors : (1407580,)\n"
     ]
    }
   ],
   "source": [
    "visitors=events[\"visitorid\"].unique()\n",
    "print('Visitor count on actions:',events[\"visitorid\"].shape)\n",
    "print('Total unique visitors :',visitors.shape)\n",
    "# unique visitors are almost half the number of total visitors"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Action Plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "sns.set(style=\"ticks\", color_codes=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.967527264504411"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "2661325 / (2661325 + 68959 + 20362) # % of views"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7720356914947213"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    " 68959 / (68959 + 20362) # % of add to carts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axs = plt.subplots(ncols=2,figsize=(15, 8))\n",
    "events_count=events[\"event\"].value_counts() # include view\n",
    "plt.title('96% of events are views')\n",
    "sns.barplot(events_count.index, events_count.values, ax=axs[0])\n",
    "events_count=events[\"event\"].value_counts()[1:] # exclude view\n",
    "plt.title('77% of non-views are add to carts')\n",
    "g=sns.barplot(events_count.index, events_count.values,ax=axs[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "data = events.event.value_counts()\n",
    "labels = data.index\n",
    "sizes = data.values\n",
    "explode = (0, 0.05, 0.15)\n",
    "fig, ax = plt.subplots(figsize=(8,8))\n",
    "\n",
    "patches, texts, autotexts = ax.pie(sizes, labels=labels, explode=explode, autopct='%1.2f%%', shadow=False, startangle=0) \n",
    "\n",
    "ax.axis('equal')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {},
   "outputs": [],
   "source": [
    "grouped = events.groupby(['event','itemid']).count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "grouped = grouped.sort_values(by=['timestamp'],ascending=False)\n",
    "grouped = grouped.reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x241725908>"
      ]
     },
     "execution_count": 162,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Views (most viewed item)\n",
    "y = grouped[grouped['event']=='view'].timestamp[0:5]\n",
    "x = grouped[grouped['event']=='view'].itemid[0:5]\n",
    "sns.barplot(x,y,order=x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x241779f98>"
      ]
     },
     "execution_count": 163,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Addtocart (most popular added to cart items)\n",
    "y = grouped[grouped['event']=='addtocart'].timestamp[0:5]\n",
    "x = grouped[grouped['event']=='addtocart'].itemid[0:5]\n",
    "sns.barplot(x,y,order=x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2419f0518>"
      ]
     },
     "execution_count": 164,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Transaction (most popular items bought)\n",
    "y = grouped[grouped['event']=='transaction'].timestamp[0:5]\n",
    "x = grouped[grouped['event']=='transaction'].itemid[0:5]\n",
    "sns.barplot(x,y,order=x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    234838.000000\n",
       "mean         11.345319\n",
       "std          35.409345\n",
       "min           1.000000\n",
       "25%           1.000000\n",
       "50%           3.000000\n",
       "75%           9.000000\n",
       "max        3410.000000\n",
       "Name: timestamp, dtype: float64"
      ]
     },
     "execution_count": 175,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# average number of views for each item is 11, \n",
    "# upper 25th percentile item has 9 views or more\n",
    "grouped[grouped.event == 'view'].timestamp.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    23903.000000\n",
       "mean         2.900556\n",
       "std          5.026141\n",
       "min          1.000000\n",
       "25%          1.000000\n",
       "50%          2.000000\n",
       "75%          3.000000\n",
       "max        306.000000\n",
       "Name: timestamp, dtype: float64"
      ]
     },
     "execution_count": 176,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# average number of views for each item is 2.9, \n",
    "# upper 25th percentile item has 3 views or more\n",
    "grouped[grouped.event == 'addtocart'].timestamp.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    12025.000000\n",
       "mean         1.867526\n",
       "std          2.710560\n",
       "min          1.000000\n",
       "25%          1.000000\n",
       "50%          1.000000\n",
       "75%          2.000000\n",
       "max        133.000000\n",
       "Name: timestamp, dtype: float64"
      ]
     },
     "execution_count": 177,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# average number of views for each item is 1.8, \n",
    "# upper 25th percentile item has 2.0 views or more\n",
    "grouped[grouped.event == 'transaction'].timestamp.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x218dffb38>"
      ]
     },
     "execution_count": 178,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# doesn't tell me anything\n",
    "corr = events[events.columns].corr()\n",
    "sns.heatmap(corr,annot = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Log-Likelihood Ratio Recommendation System\n",
    "https://towardsdatascience.com/how-to-create-a-production-ready-recommender-system-3c932752f8ea"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1407580 466866\n"
     ]
    }
   ],
   "source": [
    "n_users = events['visitorid'].unique().shape[0]\n",
    "n_items = item_prop['itemid'].max()\n",
    "print (str(n_users) +\" \" +  str(n_items))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2300.02935498"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import sys\n",
    "1407580 * 466866 * sys.getsizeof(np.int32()) / (8 * (10 ** 12)) # user-item matrix will be 2.6 TB large, let's not make this"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Because this dataset is so large, let's look at transactions only"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "trans_events = events[events.event == 'transaction']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 11719 unique visitors and 12025 unique items\n"
     ]
    }
   ],
   "source": [
    "visitors = trans_events.visitorid.unique()\n",
    "items = trans_events.itemid.unique()\n",
    "\n",
    "print(\"There are {0} unique visitors and {1} unique items\".format(visitors.shape[0], items.shape[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/meganyow/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:2: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \n",
      "/Users/meganyow/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:3: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n"
     ]
    },
    {
     "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>timestamp</th>\n",
       "      <th>visitorid</th>\n",
       "      <th>event</th>\n",
       "      <th>itemid</th>\n",
       "      <th>transactionid</th>\n",
       "      <th>visitors</th>\n",
       "      <th>items</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>130</th>\n",
       "      <td>1433222276276</td>\n",
       "      <td>599528</td>\n",
       "      <td>transaction</td>\n",
       "      <td>356475</td>\n",
       "      <td>4000.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>304</th>\n",
       "      <td>1433193500981</td>\n",
       "      <td>121688</td>\n",
       "      <td>transaction</td>\n",
       "      <td>15335</td>\n",
       "      <td>11117.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>418</th>\n",
       "      <td>1433193915008</td>\n",
       "      <td>552148</td>\n",
       "      <td>transaction</td>\n",
       "      <td>81345</td>\n",
       "      <td>5444.0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>814</th>\n",
       "      <td>1433176736375</td>\n",
       "      <td>102019</td>\n",
       "      <td>transaction</td>\n",
       "      <td>150318</td>\n",
       "      <td>13556.0</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>843</th>\n",
       "      <td>1433174518180</td>\n",
       "      <td>189384</td>\n",
       "      <td>transaction</td>\n",
       "      <td>310791</td>\n",
       "      <td>7244.0</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1214</th>\n",
       "      <td>1433184261340</td>\n",
       "      <td>350566</td>\n",
       "      <td>transaction</td>\n",
       "      <td>54058</td>\n",
       "      <td>8796.0</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1215</th>\n",
       "      <td>1433184261371</td>\n",
       "      <td>350566</td>\n",
       "      <td>transaction</td>\n",
       "      <td>284871</td>\n",
       "      <td>8796.0</td>\n",
       "      <td>5</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1234</th>\n",
       "      <td>1433182772710</td>\n",
       "      <td>404403</td>\n",
       "      <td>transaction</td>\n",
       "      <td>150100</td>\n",
       "      <td>5216.0</td>\n",
       "      <td>6</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1350</th>\n",
       "      <td>1433164275232</td>\n",
       "      <td>505565</td>\n",
       "      <td>transaction</td>\n",
       "      <td>243566</td>\n",
       "      <td>11713.0</td>\n",
       "      <td>7</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1512</th>\n",
       "      <td>1433180403455</td>\n",
       "      <td>945184</td>\n",
       "      <td>transaction</td>\n",
       "      <td>245400</td>\n",
       "      <td>2415.0</td>\n",
       "      <td>8</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1690</th>\n",
       "      <td>1433177808042</td>\n",
       "      <td>1406787</td>\n",
       "      <td>transaction</td>\n",
       "      <td>336832</td>\n",
       "      <td>15915.0</td>\n",
       "      <td>9</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1711</th>\n",
       "      <td>1433179768674</td>\n",
       "      <td>820159</td>\n",
       "      <td>transaction</td>\n",
       "      <td>202052</td>\n",
       "      <td>14924.0</td>\n",
       "      <td>10</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1871</th>\n",
       "      <td>1433202096802</td>\n",
       "      <td>319880</td>\n",
       "      <td>transaction</td>\n",
       "      <td>7943</td>\n",
       "      <td>11615.0</td>\n",
       "      <td>11</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2075</th>\n",
       "      <td>1433199718976</td>\n",
       "      <td>608628</td>\n",
       "      <td>transaction</td>\n",
       "      <td>185598</td>\n",
       "      <td>11956.0</td>\n",
       "      <td>12</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2142</th>\n",
       "      <td>1433212286455</td>\n",
       "      <td>956011</td>\n",
       "      <td>transaction</td>\n",
       "      <td>338308</td>\n",
       "      <td>3051.0</td>\n",
       "      <td>13</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2237</th>\n",
       "      <td>1433208558795</td>\n",
       "      <td>1318394</td>\n",
       "      <td>transaction</td>\n",
       "      <td>318333</td>\n",
       "      <td>11413.0</td>\n",
       "      <td>14</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2242</th>\n",
       "      <td>1433208587406</td>\n",
       "      <td>61806</td>\n",
       "      <td>transaction</td>\n",
       "      <td>68242</td>\n",
       "      <td>1343.0</td>\n",
       "      <td>15</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2292</th>\n",
       "      <td>1433209368340</td>\n",
       "      <td>1331125</td>\n",
       "      <td>transaction</td>\n",
       "      <td>185896</td>\n",
       "      <td>11138.0</td>\n",
       "      <td>16</td>\n",
       "      <td>17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2315</th>\n",
       "      <td>1433211518562</td>\n",
       "      <td>911093</td>\n",
       "      <td>transaction</td>\n",
       "      <td>277119</td>\n",
       "      <td>1798.0</td>\n",
       "      <td>17</td>\n",
       "      <td>18</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2316</th>\n",
       "      <td>1433211518578</td>\n",
       "      <td>911093</td>\n",
       "      <td>transaction</td>\n",
       "      <td>251130</td>\n",
       "      <td>1798.0</td>\n",
       "      <td>17</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2578</th>\n",
       "      <td>1433143128864</td>\n",
       "      <td>157419</td>\n",
       "      <td>transaction</td>\n",
       "      <td>63769</td>\n",
       "      <td>16605.0</td>\n",
       "      <td>18</td>\n",
       "      <td>20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2608</th>\n",
       "      <td>1433143128880</td>\n",
       "      <td>157419</td>\n",
       "      <td>transaction</td>\n",
       "      <td>135703</td>\n",
       "      <td>16605.0</td>\n",
       "      <td>18</td>\n",
       "      <td>21</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2789</th>\n",
       "      <td>1433154271395</td>\n",
       "      <td>1166562</td>\n",
       "      <td>transaction</td>\n",
       "      <td>103311</td>\n",
       "      <td>16717.0</td>\n",
       "      <td>19</td>\n",
       "      <td>22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2946</th>\n",
       "      <td>1433196261122</td>\n",
       "      <td>464138</td>\n",
       "      <td>transaction</td>\n",
       "      <td>428257</td>\n",
       "      <td>16951.0</td>\n",
       "      <td>20</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2981</th>\n",
       "      <td>1433196272991</td>\n",
       "      <td>273406</td>\n",
       "      <td>transaction</td>\n",
       "      <td>256146</td>\n",
       "      <td>8900.0</td>\n",
       "      <td>21</td>\n",
       "      <td>24</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3341</th>\n",
       "      <td>1433216459038</td>\n",
       "      <td>1233140</td>\n",
       "      <td>transaction</td>\n",
       "      <td>167827</td>\n",
       "      <td>15894.0</td>\n",
       "      <td>22</td>\n",
       "      <td>25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3346</th>\n",
       "      <td>1433216459038</td>\n",
       "      <td>1233140</td>\n",
       "      <td>transaction</td>\n",
       "      <td>387504</td>\n",
       "      <td>15894.0</td>\n",
       "      <td>22</td>\n",
       "      <td>26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3489</th>\n",
       "      <td>1433207202981</td>\n",
       "      <td>813069</td>\n",
       "      <td>transaction</td>\n",
       "      <td>201598</td>\n",
       "      <td>11179.0</td>\n",
       "      <td>23</td>\n",
       "      <td>27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3521</th>\n",
       "      <td>1433207298898</td>\n",
       "      <td>504901</td>\n",
       "      <td>transaction</td>\n",
       "      <td>367664</td>\n",
       "      <td>8467.0</td>\n",
       "      <td>24</td>\n",
       "      <td>28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3568</th>\n",
       "      <td>1433205161452</td>\n",
       "      <td>1041255</td>\n",
       "      <td>transaction</td>\n",
       "      <td>10034</td>\n",
       "      <td>17128.0</td>\n",
       "      <td>25</td>\n",
       "      <td>29</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2750509</th>\n",
       "      <td>1438368812585</td>\n",
       "      <td>1214988</td>\n",
       "      <td>transaction</td>\n",
       "      <td>464751</td>\n",
       "      <td>9316.0</td>\n",
       "      <td>11655</td>\n",
       "      <td>4059</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2750559</th>\n",
       "      <td>1438370832344</td>\n",
       "      <td>1253966</td>\n",
       "      <td>transaction</td>\n",
       "      <td>387836</td>\n",
       "      <td>4546.0</td>\n",
       "      <td>11701</td>\n",
       "      <td>11936</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2750594</th>\n",
       "      <td>1438372700741</td>\n",
       "      <td>257726</td>\n",
       "      <td>transaction</td>\n",
       "      <td>57841</td>\n",
       "      <td>3567.0</td>\n",
       "      <td>11656</td>\n",
       "      <td>4244</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2750692</th>\n",
       "      <td>1438373706127</td>\n",
       "      <td>32596</td>\n",
       "      <td>transaction</td>\n",
       "      <td>430113</td>\n",
       "      <td>8237.0</td>\n",
       "      <td>11702</td>\n",
       "      <td>10796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2750802</th>\n",
       "      <td>1438375621567</td>\n",
       "      <td>1288872</td>\n",
       "      <td>transaction</td>\n",
       "      <td>17322</td>\n",
       "      <td>1015.0</td>\n",
       "      <td>11680</td>\n",
       "      <td>9032</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2750819</th>\n",
       "      <td>1438373051591</td>\n",
       "      <td>218648</td>\n",
       "      <td>transaction</td>\n",
       "      <td>234255</td>\n",
       "      <td>1413.0</td>\n",
       "      <td>3646</td>\n",
       "      <td>220</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2750976</th>\n",
       "      <td>1438364486799</td>\n",
       "      <td>130294</td>\n",
       "      <td>transaction</td>\n",
       "      <td>426657</td>\n",
       "      <td>4931.0</td>\n",
       "      <td>11703</td>\n",
       "      <td>12013</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2750989</th>\n",
       "      <td>1438367292609</td>\n",
       "      <td>1097216</td>\n",
       "      <td>transaction</td>\n",
       "      <td>151668</td>\n",
       "      <td>15877.0</td>\n",
       "      <td>11684</td>\n",
       "      <td>12014</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2750990</th>\n",
       "      <td>1438367292594</td>\n",
       "      <td>1097216</td>\n",
       "      <td>transaction</td>\n",
       "      <td>109308</td>\n",
       "      <td>15877.0</td>\n",
       "      <td>11684</td>\n",
       "      <td>3496</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2751297</th>\n",
       "      <td>1438324747708</td>\n",
       "      <td>312976</td>\n",
       "      <td>transaction</td>\n",
       "      <td>325384</td>\n",
       "      <td>1665.0</td>\n",
       "      <td>11704</td>\n",
       "      <td>12016</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2751363</th>\n",
       "      <td>1438362585920</td>\n",
       "      <td>283269</td>\n",
       "      <td>transaction</td>\n",
       "      <td>439734</td>\n",
       "      <td>12533.0</td>\n",
       "      <td>11705</td>\n",
       "      <td>12017</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2751419</th>\n",
       "      <td>1438364101606</td>\n",
       "      <td>896088</td>\n",
       "      <td>transaction</td>\n",
       "      <td>153830</td>\n",
       "      <td>14165.0</td>\n",
       "      <td>11650</td>\n",
       "      <td>6408</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2751519</th>\n",
       "      <td>1438363115138</td>\n",
       "      <td>1118015</td>\n",
       "      <td>transaction</td>\n",
       "      <td>268883</td>\n",
       "      <td>13066.0</td>\n",
       "      <td>11706</td>\n",
       "      <td>674</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2751646</th>\n",
       "      <td>1438359544279</td>\n",
       "      <td>1163417</td>\n",
       "      <td>transaction</td>\n",
       "      <td>449776</td>\n",
       "      <td>4621.0</td>\n",
       "      <td>11707</td>\n",
       "      <td>3653</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2751697</th>\n",
       "      <td>1438357998722</td>\n",
       "      <td>1316429</td>\n",
       "      <td>transaction</td>\n",
       "      <td>14731</td>\n",
       "      <td>11423.0</td>\n",
       "      <td>11708</td>\n",
       "      <td>12018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2751721</th>\n",
       "      <td>1438359808126</td>\n",
       "      <td>1186376</td>\n",
       "      <td>transaction</td>\n",
       "      <td>132074</td>\n",
       "      <td>162.0</td>\n",
       "      <td>11709</td>\n",
       "      <td>12019</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2751759</th>\n",
       "      <td>1438359288212</td>\n",
       "      <td>104866</td>\n",
       "      <td>transaction</td>\n",
       "      <td>272969</td>\n",
       "      <td>9603.0</td>\n",
       "      <td>11710</td>\n",
       "      <td>1330</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2752444</th>\n",
       "      <td>1438394317466</td>\n",
       "      <td>118200</td>\n",
       "      <td>transaction</td>\n",
       "      <td>174923</td>\n",
       "      <td>17186.0</td>\n",
       "      <td>11711</td>\n",
       "      <td>4184</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2753046</th>\n",
       "      <td>1438361135960</td>\n",
       "      <td>331807</td>\n",
       "      <td>transaction</td>\n",
       "      <td>248455</td>\n",
       "      <td>16405.0</td>\n",
       "      <td>11712</td>\n",
       "      <td>3595</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2753612</th>\n",
       "      <td>1438394077271</td>\n",
       "      <td>722657</td>\n",
       "      <td>transaction</td>\n",
       "      <td>419980</td>\n",
       "      <td>2354.0</td>\n",
       "      <td>11696</td>\n",
       "      <td>12021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2753748</th>\n",
       "      <td>1438375193522</td>\n",
       "      <td>1099749</td>\n",
       "      <td>transaction</td>\n",
       "      <td>160671</td>\n",
       "      <td>1089.0</td>\n",
       "      <td>11682</td>\n",
       "      <td>12022</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2753826</th>\n",
       "      <td>1438375017107</td>\n",
       "      <td>899006</td>\n",
       "      <td>transaction</td>\n",
       "      <td>384823</td>\n",
       "      <td>8122.0</td>\n",
       "      <td>11713</td>\n",
       "      <td>12023</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2754521</th>\n",
       "      <td>1438385463811</td>\n",
       "      <td>1134909</td>\n",
       "      <td>transaction</td>\n",
       "      <td>20968</td>\n",
       "      <td>16316.0</td>\n",
       "      <td>11714</td>\n",
       "      <td>3693</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2754851</th>\n",
       "      <td>1438369131662</td>\n",
       "      <td>370681</td>\n",
       "      <td>transaction</td>\n",
       "      <td>178274</td>\n",
       "      <td>10966.0</td>\n",
       "      <td>11693</td>\n",
       "      <td>1062</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2754886</th>\n",
       "      <td>1438370798050</td>\n",
       "      <td>972757</td>\n",
       "      <td>transaction</td>\n",
       "      <td>418693</td>\n",
       "      <td>1702.0</td>\n",
       "      <td>11715</td>\n",
       "      <td>5690</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2755082</th>\n",
       "      <td>1438388436295</td>\n",
       "      <td>1155978</td>\n",
       "      <td>transaction</td>\n",
       "      <td>430050</td>\n",
       "      <td>4316.0</td>\n",
       "      <td>11716</td>\n",
       "      <td>6280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2755285</th>\n",
       "      <td>1438380441389</td>\n",
       "      <td>218648</td>\n",
       "      <td>transaction</td>\n",
       "      <td>446271</td>\n",
       "      <td>10485.0</td>\n",
       "      <td>3646</td>\n",
       "      <td>12024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2755294</th>\n",
       "      <td>1438377176570</td>\n",
       "      <td>1050575</td>\n",
       "      <td>transaction</td>\n",
       "      <td>31640</td>\n",
       "      <td>8354.0</td>\n",
       "      <td>11717</td>\n",
       "      <td>3246</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2755508</th>\n",
       "      <td>1438357730123</td>\n",
       "      <td>855941</td>\n",
       "      <td>transaction</td>\n",
       "      <td>235771</td>\n",
       "      <td>4385.0</td>\n",
       "      <td>11718</td>\n",
       "      <td>2419</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2755607</th>\n",
       "      <td>1438358989163</td>\n",
       "      <td>1051054</td>\n",
       "      <td>transaction</td>\n",
       "      <td>312728</td>\n",
       "      <td>17579.0</td>\n",
       "      <td>11659</td>\n",
       "      <td>188</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>19939 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             timestamp  visitorid        event  itemid  transactionid  \\\n",
       "130      1433222276276     599528  transaction  356475         4000.0   \n",
       "304      1433193500981     121688  transaction   15335        11117.0   \n",
       "418      1433193915008     552148  transaction   81345         5444.0   \n",
       "814      1433176736375     102019  transaction  150318        13556.0   \n",
       "843      1433174518180     189384  transaction  310791         7244.0   \n",
       "1214     1433184261340     350566  transaction   54058         8796.0   \n",
       "1215     1433184261371     350566  transaction  284871         8796.0   \n",
       "1234     1433182772710     404403  transaction  150100         5216.0   \n",
       "1350     1433164275232     505565  transaction  243566        11713.0   \n",
       "1512     1433180403455     945184  transaction  245400         2415.0   \n",
       "1690     1433177808042    1406787  transaction  336832        15915.0   \n",
       "1711     1433179768674     820159  transaction  202052        14924.0   \n",
       "1871     1433202096802     319880  transaction    7943        11615.0   \n",
       "2075     1433199718976     608628  transaction  185598        11956.0   \n",
       "2142     1433212286455     956011  transaction  338308         3051.0   \n",
       "2237     1433208558795    1318394  transaction  318333        11413.0   \n",
       "2242     1433208587406      61806  transaction   68242         1343.0   \n",
       "2292     1433209368340    1331125  transaction  185896        11138.0   \n",
       "2315     1433211518562     911093  transaction  277119         1798.0   \n",
       "2316     1433211518578     911093  transaction  251130         1798.0   \n",
       "2578     1433143128864     157419  transaction   63769        16605.0   \n",
       "2608     1433143128880     157419  transaction  135703        16605.0   \n",
       "2789     1433154271395    1166562  transaction  103311        16717.0   \n",
       "2946     1433196261122     464138  transaction  428257        16951.0   \n",
       "2981     1433196272991     273406  transaction  256146         8900.0   \n",
       "3341     1433216459038    1233140  transaction  167827        15894.0   \n",
       "3346     1433216459038    1233140  transaction  387504        15894.0   \n",
       "3489     1433207202981     813069  transaction  201598        11179.0   \n",
       "3521     1433207298898     504901  transaction  367664         8467.0   \n",
       "3568     1433205161452    1041255  transaction   10034        17128.0   \n",
       "...                ...        ...          ...     ...            ...   \n",
       "2750509  1438368812585    1214988  transaction  464751         9316.0   \n",
       "2750559  1438370832344    1253966  transaction  387836         4546.0   \n",
       "2750594  1438372700741     257726  transaction   57841         3567.0   \n",
       "2750692  1438373706127      32596  transaction  430113         8237.0   \n",
       "2750802  1438375621567    1288872  transaction   17322         1015.0   \n",
       "2750819  1438373051591     218648  transaction  234255         1413.0   \n",
       "2750976  1438364486799     130294  transaction  426657         4931.0   \n",
       "2750989  1438367292609    1097216  transaction  151668        15877.0   \n",
       "2750990  1438367292594    1097216  transaction  109308        15877.0   \n",
       "2751297  1438324747708     312976  transaction  325384         1665.0   \n",
       "2751363  1438362585920     283269  transaction  439734        12533.0   \n",
       "2751419  1438364101606     896088  transaction  153830        14165.0   \n",
       "2751519  1438363115138    1118015  transaction  268883        13066.0   \n",
       "2751646  1438359544279    1163417  transaction  449776         4621.0   \n",
       "2751697  1438357998722    1316429  transaction   14731        11423.0   \n",
       "2751721  1438359808126    1186376  transaction  132074          162.0   \n",
       "2751759  1438359288212     104866  transaction  272969         9603.0   \n",
       "2752444  1438394317466     118200  transaction  174923        17186.0   \n",
       "2753046  1438361135960     331807  transaction  248455        16405.0   \n",
       "2753612  1438394077271     722657  transaction  419980         2354.0   \n",
       "2753748  1438375193522    1099749  transaction  160671         1089.0   \n",
       "2753826  1438375017107     899006  transaction  384823         8122.0   \n",
       "2754521  1438385463811    1134909  transaction   20968        16316.0   \n",
       "2754851  1438369131662     370681  transaction  178274        10966.0   \n",
       "2754886  1438370798050     972757  transaction  418693         1702.0   \n",
       "2755082  1438388436295    1155978  transaction  430050         4316.0   \n",
       "2755285  1438380441389     218648  transaction  446271        10485.0   \n",
       "2755294  1438377176570    1050575  transaction   31640         8354.0   \n",
       "2755508  1438357730123     855941  transaction  235771         4385.0   \n",
       "2755607  1438358989163    1051054  transaction  312728        17579.0   \n",
       "\n",
       "         visitors  items  \n",
       "130             0      0  \n",
       "304             1      1  \n",
       "418             2      2  \n",
       "814             3      3  \n",
       "843             4      4  \n",
       "1214            5      5  \n",
       "1215            5      6  \n",
       "1234            6      7  \n",
       "1350            7      8  \n",
       "1512            8      9  \n",
       "1690            9     10  \n",
       "1711           10     11  \n",
       "1871           11     12  \n",
       "2075           12     13  \n",
       "2142           13     14  \n",
       "2237           14     15  \n",
       "2242           15     16  \n",
       "2292           16     17  \n",
       "2315           17     18  \n",
       "2316           17     19  \n",
       "2578           18     20  \n",
       "2608           18     21  \n",
       "2789           19     22  \n",
       "2946           20     23  \n",
       "2981           21     24  \n",
       "3341           22     25  \n",
       "3346           22     26  \n",
       "3489           23     27  \n",
       "3521           24     28  \n",
       "3568           25     29  \n",
       "...           ...    ...  \n",
       "2750509     11655   4059  \n",
       "2750559     11701  11936  \n",
       "2750594     11656   4244  \n",
       "2750692     11702  10796  \n",
       "2750802     11680   9032  \n",
       "2750819      3646    220  \n",
       "2750976     11703  12013  \n",
       "2750989     11684  12014  \n",
       "2750990     11684   3496  \n",
       "2751297     11704  12016  \n",
       "2751363     11705  12017  \n",
       "2751419     11650   6408  \n",
       "2751519     11706    674  \n",
       "2751646     11707   3653  \n",
       "2751697     11708  12018  \n",
       "2751721     11709  12019  \n",
       "2751759     11710   1330  \n",
       "2752444     11711   4184  \n",
       "2753046     11712   3595  \n",
       "2753612     11696  12021  \n",
       "2753748     11682  12022  \n",
       "2753826     11713  12023  \n",
       "2754521     11714   3693  \n",
       "2754851     11693   1062  \n",
       "2754886     11715   5690  \n",
       "2755082     11716   6280  \n",
       "2755285      3646  12024  \n",
       "2755294     11717   3246  \n",
       "2755508     11718   2419  \n",
       "2755607     11659    188  \n",
       "\n",
       "[19939 rows x 7 columns]"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trans2 = trans_events.groupby('visitorid').head(50)\n",
    "trans2['visitors'] = trans2['visitorid'].apply(lambda x : np.argwhere(visitors == x)[0][0])\n",
    "trans2['items'] = trans2['itemid'].apply(lambda x : np.argwhere(items == x)[0][0])\n",
    "trans2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "493.2234125"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "11719 * 12025 * sys.getsizeof(np.int32()) / (8 * (10 ** 6)) \n",
    "# user-item matrix will be 500MB large, sparse matrix to the rescue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.sparse import csr_matrix, lil_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<11719x12025 sparse matrix of type '<class 'numpy.int8'>'\n",
       "\twith 18905 stored elements in LInked List format>"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "occurences = lil_matrix((visitors.shape[0], items.shape[0]), dtype='int8')\n",
    "\n",
    "def set_occurences(visitor, item):\n",
    "    occurences[visitor, item] += 1\n",
    "    \n",
    "trans2.apply(lambda row: set_occurences(row['visitors'], row['items']), axis=1)\n",
    "occurences"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From those 140 million cells in the matrix, only 18,905 are filled with non-zero.  \n",
    "So basically you only need to store those 18,905 value to the memory. A 99.99% improved efficiency.  \n",
    "The downside of the sparse matrix is, it is computationally higher when trying to retrieve the data in real-time. So, you should not finish at this step.  \n",
    "\n",
    "Now let's create a co-occurence matrix i.e. an item-item matrix where each element means how many times both items bought togehter by a user.  \n",
    "To create one we need to do the dot product the transpose of the occursence matrix with itself.  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/meganyow/anaconda3/lib/python3.7/site-packages/scipy/sparse/compressed.py:746: SparseEfficiencyWarning: Changing the sparsity structure of a csr_matrix is expensive. lil_matrix is more efficient.\n",
      "  SparseEfficiencyWarning)\n"
     ]
    }
   ],
   "source": [
    "cooc = occurences.transpose().dot(occurences)\n",
    "cooc.setdiag(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Use Log-Likelihood Ratio to penalize the score of co-occurence matrix to make predictions generalizable  \n",
    "remove effects on purchasing behaviour caused by flash sale  \n",
    "http://tdunning.blogspot.com/2008/03/surprise-and-coincidence.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "def xLogX(x):\n",
    "    return x * np.log(x) if x != 0 else 0.0\n",
    "\n",
    "def entropy(x1, x2=0, x3=0, x4=0):\n",
    "    return xLogX(x1 + x2 + x3 + x4) - xLogX(x1) - xLogX(x2) - xLogX(x3) - xLogX(x4)\n",
    "\n",
    "def LLR(k11, k12, k21, k22):\n",
    "    \"\"\"\n",
    "    calculate likelihood of two events, A and B appear together\n",
    "\n",
    "    k11, number of when both events appeared together\n",
    "    k12, number of B appear without A\n",
    "    k21, number of A appear without B\n",
    "    k22, number of other things appeared without both of them\n",
    "    \"\"\"\n",
    "    rowEntropy = entropy(k11 + k12, k21 + k22)\n",
    "    columnEntropy = entropy(k11 + k21, k12 + k22)\n",
    "    matrixEntropy = entropy(k11, k12, k21, k22)\n",
    "    if rowEntropy + columnEntropy < matrixEntropy:\n",
    "        return 0.0\n",
    "    return 2.0 * (rowEntropy + columnEntropy - matrixEntropy)\n",
    "\n",
    "def rootLLR(k11, k12, k21, k22):\n",
    "    llr = LLR(k11, k12, k21, k22)\n",
    "    sqrt = np.sqrt(llr)\n",
    "    if k11 * 1.0 / (k11 + k12) < k21 * 1.0 / (k21 + k22):\n",
    "        sqrt = -sqrt\n",
    "    return sqrt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/meganyow/anaconda3/lib/python3.7/site-packages/scipy/sparse/compressed.py:746: SparseEfficiencyWarning: Changing the sparsity structure of a csr_matrix is expensive. lil_matrix is more efficient.\n",
      "  SparseEfficiencyWarning)\n"
     ]
    }
   ],
   "source": [
    "row_sum = np.sum(cooc, axis=0).A.flatten()\n",
    "column_sum = np.sum(cooc, axis=1).A.flatten()\n",
    "total = np.sum(row_sum, axis=0)\n",
    "pp_score = csr_matrix((cooc.shape[0], cooc.shape[1]), dtype='double')\n",
    "cx = cooc.tocoo()\n",
    "\n",
    "for i,j,v in zip(cx.row, cx.col, cx.data):\n",
    "    if v != 0:\n",
    "        k11 = v\n",
    "        k12 = row_sum[i] - k11\n",
    "        k21 = column_sum[j] - k11\n",
    "        k22 = total - k11 - k12 - k21\n",
    "        pp_score[i,j] = rootLLR(k11, k12, k21, k22)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "result = np.flip(np.sort(pp_score.A, axis=1), axis=1)\n",
    "result_indices = np.flip(np.argsort(pp_score.A, axis=1), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([15.33511076, 14.60017668,  3.62091635, ...,  0.        ,\n",
       "        0.        ,  0.        ])"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result[8456]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([8682,  380, 8501, ..., 8010, 8009,    0])"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result_indices[8456]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can safely answer that with a high number of LLR score, item 8682 and 380 can be an indicator for item 8456. While item 8501 since the score is not that big, might not be an indicator for item 8456.\n",
    "It means that, if someone bought 8682 and 380, you can recommend him 8456."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "minLLR = 5\n",
    "indicators = result[:, :50]\n",
    "indicators[indicators < minLLR] = 0.0\n",
    "indicators_indices = result_indices[:, :50]\n",
    "max_indicator_indices = (indicators==0).argmax(axis=1)\n",
    "max = max_indicator_indices.max()\n",
    "indicators = indicators[:, :max+1]\n",
    "indicators_indices = indicators_indices[:, :max+1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Productionalize Model\n",
    "\n",
    "Using elasticsearch using bulk API that can send multiple documents at once"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "import requests\n",
    "import json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "def bulk_update(url, items):\n",
    "    actions = []\n",
    "    for i in range(indicators.shape[0]):\n",
    "        length = indicators[i].nonzero()[0].shape[0]\n",
    "        real_indicators = items[indicators_indices[i, :length]].astype(\"int\").tolist()\n",
    "        id = items[i]\n",
    "\n",
    "        action = { \"index\" : { \"_index\" : \"items2\", \"_id\" : str(id) } }\n",
    "\n",
    "        data = {\n",
    "            \"id\": int(id),\n",
    "            \"indicators\": real_indicators\n",
    "        }\n",
    "\n",
    "        actions.append(json.dumps(action))\n",
    "        actions.append(json.dumps(data))\n",
    "\n",
    "        if len(actions) == 200:\n",
    "            actions_string = \"\\n\".join(actions) + \"\\n\"\n",
    "            actions = []\n",
    "\n",
    "            #url = \"http://127.0.0.1:9200/_bulk/\"\n",
    "            headers = {\n",
    "                \"Content-Type\" : \"application/x-ndjson\"\n",
    "            }\n",
    "            requests.post(url, headers=headers, data=actions_string)\n",
    "    if len(actions) > 0:\n",
    "        actions_string = \"\\n\".join(actions) + \"\\n\"\n",
    "        actions = []\n",
    "        #url = \"http://127.0.0.1:9200/_bulk/\"\n",
    "        headers = {\n",
    "            \"Content-Type\" : \"application/x-ndjson\"\n",
    "        }\n",
    "        requests.post(url, headers=headers, data=actions_string)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}