671_project.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import json\n",
    "import gzip\n",
    "import datetime, time\n",
    "\n",
    "import pickle\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Read and Inspect Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def parse(path):\n",
    "  g = gzip.open(path, 'rb')\n",
    "  for l in g:\n",
    "    yield json.loads(l)\n",
    "\n",
    "def getDF(path):\n",
    "  i = 0\n",
    "  df = {}\n",
    "  for d in parse(path):\n",
    "    df[i] = d\n",
    "    i += 1\n",
    "  return pd.DataFrame.from_dict(df, orient='index')\n",
    "\n",
    "interaction = getDF('data/Amazon_Fashion.jsonl.gz')\n",
    "meta = getDF('data/meta_Amazon_Fashion.jsonl.gz')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((2500939, 10), (826108, 14))"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interaction.shape, meta.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>rating</th>\n",
       "      <th>title</th>\n",
       "      <th>text</th>\n",
       "      <th>images</th>\n",
       "      <th>asin</th>\n",
       "      <th>parent_asin</th>\n",
       "      <th>user_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>helpful_vote</th>\n",
       "      <th>verified_purchase</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>Pretty locket</td>\n",
       "      <td>I think this locket is really pretty. The insi...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B00LOPVX74</td>\n",
       "      <td>B00LOPVX74</td>\n",
       "      <td>AGBFYI2DDIKXC5Y4FARTYDTQBMFQ</td>\n",
       "      <td>1578528394489</td>\n",
       "      <td>3</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5.0</td>\n",
       "      <td>A</td>\n",
       "      <td>Great</td>\n",
       "      <td>[]</td>\n",
       "      <td>B07B4JXK8D</td>\n",
       "      <td>B07B4JXK8D</td>\n",
       "      <td>AFQLNQNQYFWQZPJQZS6V3NZU4QBQ</td>\n",
       "      <td>1608426246701</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>Two Stars</td>\n",
       "      <td>One of the stones fell out within the first 2 ...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B007ZSEQ4Q</td>\n",
       "      <td>B007ZSEQ4Q</td>\n",
       "      <td>AHITBJSS7KYUBVZPX7M2WJCOIVKQ</td>\n",
       "      <td>1432344828000</td>\n",
       "      <td>3</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.0</td>\n",
       "      <td>Won’t buy again</td>\n",
       "      <td>Crappy socks. Money wasted. Bought to wear wit...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B07F2BTFS9</td>\n",
       "      <td>B07F2BTFS9</td>\n",
       "      <td>AFVNEEPDEIH5SPUN5BWC6NKL3WNQ</td>\n",
       "      <td>1546289847095</td>\n",
       "      <td>2</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>I LOVE these glasses</td>\n",
       "      <td>I LOVE these glasses!  They fit perfectly over...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B00PKRFU4O</td>\n",
       "      <td>B00XESJTDE</td>\n",
       "      <td>AHSPLDNW5OOUK2PLH7GXLACFBZNQ</td>\n",
       "      <td>1439476166000</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   rating                 title  \\\n",
       "0     5.0         Pretty locket   \n",
       "1     5.0                     A   \n",
       "2     2.0             Two Stars   \n",
       "3     1.0       Won’t buy again   \n",
       "4     5.0  I LOVE these glasses   \n",
       "\n",
       "                                                text images        asin  \\\n",
       "0  I think this locket is really pretty. The insi...     []  B00LOPVX74   \n",
       "1                                              Great     []  B07B4JXK8D   \n",
       "2  One of the stones fell out within the first 2 ...     []  B007ZSEQ4Q   \n",
       "3  Crappy socks. Money wasted. Bought to wear wit...     []  B07F2BTFS9   \n",
       "4  I LOVE these glasses!  They fit perfectly over...     []  B00PKRFU4O   \n",
       "\n",
       "  parent_asin                       user_id      timestamp  helpful_vote  \\\n",
       "0  B00LOPVX74  AGBFYI2DDIKXC5Y4FARTYDTQBMFQ  1578528394489             3   \n",
       "1  B07B4JXK8D  AFQLNQNQYFWQZPJQZS6V3NZU4QBQ  1608426246701             0   \n",
       "2  B007ZSEQ4Q  AHITBJSS7KYUBVZPX7M2WJCOIVKQ  1432344828000             3   \n",
       "3  B07F2BTFS9  AFVNEEPDEIH5SPUN5BWC6NKL3WNQ  1546289847095             2   \n",
       "4  B00XESJTDE  AHSPLDNW5OOUK2PLH7GXLACFBZNQ  1439476166000             0   \n",
       "\n",
       "   verified_purchase  \n",
       "0               True  \n",
       "1               True  \n",
       "2               True  \n",
       "3               True  \n",
       "4               True  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interaction.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "rating\n",
       "5.0    1426268\n",
       "4.0     352327\n",
       "1.0     303411\n",
       "3.0     245471\n",
       "2.0     173462\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interaction.rating.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['rating', 'title', 'text', 'images', 'asin', 'parent_asin', 'user_id',\n",
       "       'timestamp', 'helpful_vote', 'verified_purchase'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interaction.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "rating               0\n",
       "title                0\n",
       "text                 0\n",
       "images               0\n",
       "asin                 0\n",
       "parent_asin          0\n",
       "user_id              0\n",
       "timestamp            0\n",
       "helpful_vote         0\n",
       "verified_purchase    0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interaction.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "verified_purchase\n",
       "True     2337702\n",
       "False     163237\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 122,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interaction.verified_purchase.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(Timestamp('2002-05-07 01:51:28'), Timestamp('2023-09-11 03:24:38.515000'))"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.to_datetime(interaction.timestamp.min(), unit='ms'), pd.to_datetime(interaction.timestamp.max(), unit='ms')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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>user_id</th>\n",
       "      <th>count_asin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>286</th>\n",
       "      <td>AE22M44KAE72RFU6DWM5Y3YN6SEA</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>611</th>\n",
       "      <td>AE23BYWB52METWQVHSPN3MKN7AJA</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>653</th>\n",
       "      <td>AE23ERACMJTGJJVEKZHU5VU6YZHA</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>963</th>\n",
       "      <td>AE23ZBUF2YVBQPH2NN6F5XSA3QYQ</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1406</th>\n",
       "      <td>AE24VPMWEEQD62YPOG53BW7JCGFA</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2034984</th>\n",
       "      <td>AHZZ2OXOLI63BPXIVEMTWFS77S7A</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2035008</th>\n",
       "      <td>AHZZ476HPE5T27CPN62OV6RAM6KQ</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2035048</th>\n",
       "      <td>AHZZ6SGWPYHMV7EKQDBVUTHHFZUA</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2035153</th>\n",
       "      <td>AHZZFHROQ52EDDBX4FOQIEPSK5ZQ</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2035201</th>\n",
       "      <td>AHZZIENTC5NQS32NSKEI7FWPDLWQ</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>14774 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                              user_id  count_asin\n",
       "286      AE22M44KAE72RFU6DWM5Y3YN6SEA           9\n",
       "611      AE23BYWB52METWQVHSPN3MKN7AJA           8\n",
       "653      AE23ERACMJTGJJVEKZHU5VU6YZHA          11\n",
       "963      AE23ZBUF2YVBQPH2NN6F5XSA3QYQ           7\n",
       "1406     AE24VPMWEEQD62YPOG53BW7JCGFA           9\n",
       "...                               ...         ...\n",
       "2034984  AHZZ2OXOLI63BPXIVEMTWFS77S7A           5\n",
       "2035008  AHZZ476HPE5T27CPN62OV6RAM6KQ           5\n",
       "2035048  AHZZ6SGWPYHMV7EKQDBVUTHHFZUA          11\n",
       "2035153  AHZZFHROQ52EDDBX4FOQIEPSK5ZQ           5\n",
       "2035201  AHZZIENTC5NQS32NSKEI7FWPDLWQ           9\n",
       "\n",
       "[14774 rows x 2 columns]"
      ]
     },
     "execution_count": 128,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "count_user = interaction.groupby('user_id').agg({'asin': 'nunique'}).reset_index().rename(columns={'asin': 'count_asin'})\n",
    "\n",
    "count_user[count_user.count_asin>=5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(14774, 2)"
      ]
     },
     "execution_count": 130,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "count_user[count_user.count_asin>=5].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "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>asin</th>\n",
       "      <th>count_user</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>0545298156</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>0599521503</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>0610318861</td>\n",
       "      <td>47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>0615991335</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>0629150591</td>\n",
       "      <td>103</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>873893</th>\n",
       "      <td>B0BX5PL8C5</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>873925</th>\n",
       "      <td>B0BXDCK1JD</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>873930</th>\n",
       "      <td>B0BXGZFLJ6</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>873944</th>\n",
       "      <td>B0BXNWBR6S</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>874112</th>\n",
       "      <td>B0BZKY1Q64</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>98999 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              asin  count_user\n",
       "14      0545298156           6\n",
       "15      0599521503          16\n",
       "16      0610318861          47\n",
       "18      0615991335          12\n",
       "19      0629150591         103\n",
       "...            ...         ...\n",
       "873893  B0BX5PL8C5          13\n",
       "873925  B0BXDCK1JD           5\n",
       "873930  B0BXGZFLJ6           7\n",
       "873944  B0BXNWBR6S           5\n",
       "874112  B0BZKY1Q64           5\n",
       "\n",
       "[98999 rows x 2 columns]"
      ]
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "count_item = interaction.groupby('asin').agg({'user_id': 'nunique'}).reset_index().rename(columns={'user_id': 'count_user'})\n",
    "\n",
    "count_item[count_item.count_user >= 5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(98999, 2)"
      ]
     },
     "execution_count": 132,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "count_item[count_item.count_user >= 5].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>main_category</th>\n",
       "      <th>title</th>\n",
       "      <th>average_rating</th>\n",
       "      <th>rating_number</th>\n",
       "      <th>features</th>\n",
       "      <th>description</th>\n",
       "      <th>price</th>\n",
       "      <th>images</th>\n",
       "      <th>videos</th>\n",
       "      <th>store</th>\n",
       "      <th>categories</th>\n",
       "      <th>details</th>\n",
       "      <th>parent_asin</th>\n",
       "      <th>bought_together</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>AMAZON FASHION</td>\n",
       "      <td>YUEDGE 5 Pairs Men's Moisture Control Cushione...</td>\n",
       "      <td>4.6</td>\n",
       "      <td>16</td>\n",
       "      <td>[]</td>\n",
       "      <td>[]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>[{'thumb': 'https://m.media-amazon.com/images/...</td>\n",
       "      <td>[]</td>\n",
       "      <td>GiveGift</td>\n",
       "      <td>[]</td>\n",
       "      <td>{'Package Dimensions': '10.31 x 8.5 x 1.73 inc...</td>\n",
       "      <td>B08BHN9PK5</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>AMAZON FASHION</td>\n",
       "      <td>DouBCQ Women's Palazzo Lounge Wide Leg Casual ...</td>\n",
       "      <td>4.1</td>\n",
       "      <td>7</td>\n",
       "      <td>[Drawstring closure, Machine Wash]</td>\n",
       "      <td>[]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>[{'thumb': 'https://m.media-amazon.com/images/...</td>\n",
       "      <td>[]</td>\n",
       "      <td>DouBCQ</td>\n",
       "      <td>[]</td>\n",
       "      <td>{'Package Dimensions': '15 x 10.2 x 0.4 inches...</td>\n",
       "      <td>B08R39MRDW</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    main_category                                              title  \\\n",
       "0  AMAZON FASHION  YUEDGE 5 Pairs Men's Moisture Control Cushione...   \n",
       "1  AMAZON FASHION  DouBCQ Women's Palazzo Lounge Wide Leg Casual ...   \n",
       "\n",
       "   average_rating  rating_number                            features  \\\n",
       "0             4.6             16                                  []   \n",
       "1             4.1              7  [Drawstring closure, Machine Wash]   \n",
       "\n",
       "  description  price                                             images  \\\n",
       "0          []    NaN  [{'thumb': 'https://m.media-amazon.com/images/...   \n",
       "1          []    NaN  [{'thumb': 'https://m.media-amazon.com/images/...   \n",
       "\n",
       "  videos     store categories  \\\n",
       "0     []  GiveGift         []   \n",
       "1     []    DouBCQ         []   \n",
       "\n",
       "                                             details parent_asin  \\\n",
       "0  {'Package Dimensions': '10.31 x 8.5 x 1.73 inc...  B08BHN9PK5   \n",
       "1  {'Package Dimensions': '15 x 10.2 x 0.4 inches...  B08R39MRDW   \n",
       "\n",
       "  bought_together  \n",
       "0            None  \n",
       "1            None  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meta.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "main_category           0\n",
       "title                   0\n",
       "average_rating          0\n",
       "rating_number           0\n",
       "features                0\n",
       "description             0\n",
       "price              775859\n",
       "images                  0\n",
       "videos                  0\n",
       "store               26838\n",
       "categories              0\n",
       "details                 0\n",
       "parent_asin             0\n",
       "bought_together    826108\n",
       "dtype: int64"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meta.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(826108, 14)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meta.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Processing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Filter interaction data\n",
    "* Keep only verified purchases  \n",
    "* Keep only records whose asin is the same as parent asin and is also in product metadata  \n",
    "* Keep only userful columns: user id, item id, time stamp, rating\n",
    "* Keep records whose user id has **reviewed at least 5 unique items** in the dataset (thus we can evaluate our model)   \n",
    "\n",
    "Note: the order of step 2 and step 4 should not be reversed!!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "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>rating</th>\n",
       "      <th>title</th>\n",
       "      <th>text</th>\n",
       "      <th>images</th>\n",
       "      <th>asin</th>\n",
       "      <th>parent_asin</th>\n",
       "      <th>user_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>helpful_vote</th>\n",
       "      <th>verified_purchase</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>Pretty locket</td>\n",
       "      <td>I think this locket is really pretty. The insi...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B00LOPVX74</td>\n",
       "      <td>B00LOPVX74</td>\n",
       "      <td>AGBFYI2DDIKXC5Y4FARTYDTQBMFQ</td>\n",
       "      <td>1578528394489</td>\n",
       "      <td>3</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   rating          title                                               text  \\\n",
       "0     5.0  Pretty locket  I think this locket is really pretty. The insi...   \n",
       "\n",
       "  images        asin parent_asin                       user_id      timestamp  \\\n",
       "0     []  B00LOPVX74  B00LOPVX74  AGBFYI2DDIKXC5Y4FARTYDTQBMFQ  1578528394489   \n",
       "\n",
       "   helpful_vote  verified_purchase  \n",
       "0             3               True  "
      ]
     },
     "execution_count": 197,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interaction.head(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2045374, 10)"
      ]
     },
     "execution_count": 198,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ratings = interaction[(interaction.asin.isin(meta.parent_asin)) & (interaction.verified_purchase == True)]\n",
    "\n",
    "## note: interaction.asin.isin(meta.parent_asin) ensures the asin is the same as parent asin and is also in product metadata\n",
    "\n",
    "ratings.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9566, 2)"
      ]
     },
     "execution_count": 199,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "count_user = ratings.groupby('user_id').agg({'asin': 'nunique'}).reset_index().rename(columns={'asin': 'count_asin'})\n",
    "\n",
    "count_user[count_user.count_asin>=5].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(63229, 10)"
      ]
     },
     "execution_count": 200,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ratings = ratings[(ratings.user_id.isin(count_user[count_user.count_asin>=5].user_id))]\n",
    "\n",
    "ratings.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "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>rating</th>\n",
       "      <th>title</th>\n",
       "      <th>text</th>\n",
       "      <th>images</th>\n",
       "      <th>asin</th>\n",
       "      <th>parent_asin</th>\n",
       "      <th>user_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>helpful_vote</th>\n",
       "      <th>verified_purchase</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>2.0</td>\n",
       "      <td>elastic adjustable bands for charge 4</td>\n",
       "      <td>Well.  I really like the idea, however, they a...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B098NYMTYH</td>\n",
       "      <td>B098NYMTYH</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>1661808780730</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49</th>\n",
       "      <td>2.0</td>\n",
       "      <td>open shoulder t shirt</td>\n",
       "      <td>I am 5'7\".. usually wear a 14/16 or 0x/1x.. 38...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B095CHR38Z</td>\n",
       "      <td>B095CHR38Z</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>1645062075013</td>\n",
       "      <td>2</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>5.0</td>\n",
       "      <td>fancy pumpkin headband</td>\n",
       "      <td>Purchased for my sister to use on Halloween......</td>\n",
       "      <td>[]</td>\n",
       "      <td>B071HMN7K8</td>\n",
       "      <td>B071HMN7K8</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>1635309000822</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>1.0</td>\n",
       "      <td>Nike Jacket?</td>\n",
       "      <td>The photo doesn't show this but you need a col...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B074XF4KG4</td>\n",
       "      <td>B074XF4KG4</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>1611092662402</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53</th>\n",
       "      <td>4.0</td>\n",
       "      <td>green turtleneck</td>\n",
       "      <td>Slimfit green turtleneck... I am 5'7\"... 38DDD...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B08KXH42H9</td>\n",
       "      <td>B08KXH42H9</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>1607128372655</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    rating                                  title  \\\n",
       "48     2.0  elastic adjustable bands for charge 4   \n",
       "49     2.0                  open shoulder t shirt   \n",
       "50     5.0                 fancy pumpkin headband   \n",
       "52     1.0                           Nike Jacket?   \n",
       "53     4.0                       green turtleneck   \n",
       "\n",
       "                                                 text images        asin  \\\n",
       "48  Well.  I really like the idea, however, they a...     []  B098NYMTYH   \n",
       "49  I am 5'7\".. usually wear a 14/16 or 0x/1x.. 38...     []  B095CHR38Z   \n",
       "50  Purchased for my sister to use on Halloween......     []  B071HMN7K8   \n",
       "52  The photo doesn't show this but you need a col...     []  B074XF4KG4   \n",
       "53  Slimfit green turtleneck... I am 5'7\"... 38DDD...     []  B08KXH42H9   \n",
       "\n",
       "   parent_asin                       user_id      timestamp  helpful_vote  \\\n",
       "48  B098NYMTYH  AFZUK3MTBIBEDQOPAK3OATUOUKLA  1661808780730             0   \n",
       "49  B095CHR38Z  AFZUK3MTBIBEDQOPAK3OATUOUKLA  1645062075013             2   \n",
       "50  B071HMN7K8  AFZUK3MTBIBEDQOPAK3OATUOUKLA  1635309000822             0   \n",
       "52  B074XF4KG4  AFZUK3MTBIBEDQOPAK3OATUOUKLA  1611092662402             0   \n",
       "53  B08KXH42H9  AFZUK3MTBIBEDQOPAK3OATUOUKLA  1607128372655             0   \n",
       "\n",
       "    verified_purchase  \n",
       "48               True  \n",
       "49               True  \n",
       "50               True  \n",
       "52               True  \n",
       "53               True  "
      ]
     },
     "execution_count": 201,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ratings.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 209,
   "metadata": {},
   "outputs": [],
   "source": [
    "## last interaction of each user\n",
    "\n",
    "loo = ratings.loc[ratings.groupby('user_id')['timestamp'].idxmax()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 210,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((53663, 10), (9566, 10), (63229, 10))"
      ]
     },
     "execution_count": 210,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## remained interactions excluding last interaction of each user\n",
    "\n",
    "remain = ratings[~ratings.index.isin(loo.index)]\n",
    "remain.shape, loo.shape, ratings.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "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>user_id</th>\n",
       "      <th>asin</th>\n",
       "      <th>rating</th>\n",
       "      <th>timestamp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1465723</th>\n",
       "      <td>AE22M44KAE72RFU6DWM5Y3YN6SEA</td>\n",
       "      <td>B01LXOXLA1</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1494008552000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1008584</th>\n",
       "      <td>AE23BYWB52METWQVHSPN3MKN7AJA</td>\n",
       "      <td>B094J98QT9</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1653146545518</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>794474</th>\n",
       "      <td>AE23ERACMJTGJJVEKZHU5VU6YZHA</td>\n",
       "      <td>B0713PN77B</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1560016417138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1481228</th>\n",
       "      <td>AE26PXEWX73ULGW5RAROBYPTWCWA</td>\n",
       "      <td>B0779ZCLG1</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1533549549679</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1690323</th>\n",
       "      <td>AE26QJAXLZIQC4UNFKQFA4M2A3VA</td>\n",
       "      <td>B07WCLP2GK</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1569034094097</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                              user_id        asin  rating      timestamp\n",
       "1465723  AE22M44KAE72RFU6DWM5Y3YN6SEA  B01LXOXLA1     3.0  1494008552000\n",
       "1008584  AE23BYWB52METWQVHSPN3MKN7AJA  B094J98QT9     1.0  1653146545518\n",
       "794474   AE23ERACMJTGJJVEKZHU5VU6YZHA  B0713PN77B     5.0  1560016417138\n",
       "1481228  AE26PXEWX73ULGW5RAROBYPTWCWA  B0779ZCLG1     5.0  1533549549679\n",
       "1690323  AE26QJAXLZIQC4UNFKQFA4M2A3VA  B07WCLP2GK     5.0  1569034094097"
      ]
     },
     "execution_count": 212,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loo[['user_id', 'asin', 'rating', 'timestamp']].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "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>user_id</th>\n",
       "      <th>asin</th>\n",
       "      <th>rating</th>\n",
       "      <th>timestamp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>794474</th>\n",
       "      <td>AE23ERACMJTGJJVEKZHU5VU6YZHA</td>\n",
       "      <td>B0713PN77B</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1560016417138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>785680</th>\n",
       "      <td>AE274ELTNGUU6B3ZF7HUUJARXUIQ</td>\n",
       "      <td>B00N9X8VHY</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1475322186000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1242387</th>\n",
       "      <td>AE2E2EBMC32XMPHHYTQBXU4MXCNA</td>\n",
       "      <td>B01EN4KVRK</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1498787164039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>945782</th>\n",
       "      <td>AE2MR33PBSOMX3ISKWRFFWP2V66Q</td>\n",
       "      <td>B00KA3SRVG</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1460551598000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>726398</th>\n",
       "      <td>AE2NF4KWRNOCX3AELFDWT3XSWSQQ</td>\n",
       "      <td>B07PR6F5JD</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1560815099430</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1540721</th>\n",
       "      <td>AHZP4XCD3RAB4KJG3J5N4QYESCKA</td>\n",
       "      <td>B00PJIPH2O</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1483042586000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>909784</th>\n",
       "      <td>AHZP5Q474KKOF465OFANSHL6FRXA</td>\n",
       "      <td>B01K8YIB8K</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1536254130657</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>570908</th>\n",
       "      <td>AHZUZ3YYBUE7D7DG5QJ46EDASOOQ</td>\n",
       "      <td>B00KATGS16</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1481438428000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>610838</th>\n",
       "      <td>AHZXRKXTXBLR5JJKFQZ3ME2VWCMA</td>\n",
       "      <td>B00KA3VX62</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1529615304471</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>109895</th>\n",
       "      <td>AHZZ6SGWPYHMV7EKQDBVUTHHFZUA</td>\n",
       "      <td>B01A9MA3ZK</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1529283444064</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1426 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                              user_id        asin  rating      timestamp\n",
       "794474   AE23ERACMJTGJJVEKZHU5VU6YZHA  B0713PN77B     5.0  1560016417138\n",
       "785680   AE274ELTNGUU6B3ZF7HUUJARXUIQ  B00N9X8VHY     5.0  1475322186000\n",
       "1242387  AE2E2EBMC32XMPHHYTQBXU4MXCNA  B01EN4KVRK     5.0  1498787164039\n",
       "945782   AE2MR33PBSOMX3ISKWRFFWP2V66Q  B00KA3SRVG     5.0  1460551598000\n",
       "726398   AE2NF4KWRNOCX3AELFDWT3XSWSQQ  B07PR6F5JD     5.0  1560815099430\n",
       "...                               ...         ...     ...            ...\n",
       "1540721  AHZP4XCD3RAB4KJG3J5N4QYESCKA  B00PJIPH2O     2.0  1483042586000\n",
       "909784   AHZP5Q474KKOF465OFANSHL6FRXA  B01K8YIB8K     5.0  1536254130657\n",
       "570908   AHZUZ3YYBUE7D7DG5QJ46EDASOOQ  B00KATGS16     5.0  1481438428000\n",
       "610838   AHZXRKXTXBLR5JJKFQZ3ME2VWCMA  B00KA3VX62     5.0  1529615304471\n",
       "109895   AHZZ6SGWPYHMV7EKQDBVUTHHFZUA  B01A9MA3ZK     4.0  1529283444064\n",
       "\n",
       "[1426 rows x 4 columns]"
      ]
     },
     "execution_count": 213,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loo[loo.asin.isin(remain.asin)][['user_id', 'asin', 'rating', 'timestamp']]\n",
    "\n",
    "## 1426 users whose last interacted item is in the remained interactions. These records can be used as test set so that the test items all appear in the training set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1426, 10)"
      ]
     },
     "execution_count": 214,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_set = loo[loo.asin.isin(remain.asin)]\n",
    "test_userid = loo[loo.asin.isin(remain.asin)].user_id\n",
    "test_set.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "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>rating</th>\n",
       "      <th>title</th>\n",
       "      <th>text</th>\n",
       "      <th>images</th>\n",
       "      <th>asin</th>\n",
       "      <th>parent_asin</th>\n",
       "      <th>user_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>helpful_vote</th>\n",
       "      <th>verified_purchase</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>2.0</td>\n",
       "      <td>elastic adjustable bands for charge 4</td>\n",
       "      <td>Well.  I really like the idea, however, they a...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B098NYMTYH</td>\n",
       "      <td>B098NYMTYH</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>1661808780730</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49</th>\n",
       "      <td>2.0</td>\n",
       "      <td>open shoulder t shirt</td>\n",
       "      <td>I am 5'7\".. usually wear a 14/16 or 0x/1x.. 38...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B095CHR38Z</td>\n",
       "      <td>B095CHR38Z</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>1645062075013</td>\n",
       "      <td>2</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>5.0</td>\n",
       "      <td>fancy pumpkin headband</td>\n",
       "      <td>Purchased for my sister to use on Halloween......</td>\n",
       "      <td>[]</td>\n",
       "      <td>B071HMN7K8</td>\n",
       "      <td>B071HMN7K8</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>1635309000822</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>1.0</td>\n",
       "      <td>Nike Jacket?</td>\n",
       "      <td>The photo doesn't show this but you need a col...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B074XF4KG4</td>\n",
       "      <td>B074XF4KG4</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>1611092662402</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53</th>\n",
       "      <td>4.0</td>\n",
       "      <td>green turtleneck</td>\n",
       "      <td>Slimfit green turtleneck... I am 5'7\"... 38DDD...</td>\n",
       "      <td>[]</td>\n",
       "      <td>B08KXH42H9</td>\n",
       "      <td>B08KXH42H9</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>1607128372655</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    rating                                  title  \\\n",
       "48     2.0  elastic adjustable bands for charge 4   \n",
       "49     2.0                  open shoulder t shirt   \n",
       "50     5.0                 fancy pumpkin headband   \n",
       "52     1.0                           Nike Jacket?   \n",
       "53     4.0                       green turtleneck   \n",
       "\n",
       "                                                 text images        asin  \\\n",
       "48  Well.  I really like the idea, however, they a...     []  B098NYMTYH   \n",
       "49  I am 5'7\".. usually wear a 14/16 or 0x/1x.. 38...     []  B095CHR38Z   \n",
       "50  Purchased for my sister to use on Halloween......     []  B071HMN7K8   \n",
       "52  The photo doesn't show this but you need a col...     []  B074XF4KG4   \n",
       "53  Slimfit green turtleneck... I am 5'7\"... 38DDD...     []  B08KXH42H9   \n",
       "\n",
       "   parent_asin                       user_id      timestamp  helpful_vote  \\\n",
       "48  B098NYMTYH  AFZUK3MTBIBEDQOPAK3OATUOUKLA  1661808780730             0   \n",
       "49  B095CHR38Z  AFZUK3MTBIBEDQOPAK3OATUOUKLA  1645062075013             2   \n",
       "50  B071HMN7K8  AFZUK3MTBIBEDQOPAK3OATUOUKLA  1635309000822             0   \n",
       "52  B074XF4KG4  AFZUK3MTBIBEDQOPAK3OATUOUKLA  1611092662402             0   \n",
       "53  B08KXH42H9  AFZUK3MTBIBEDQOPAK3OATUOUKLA  1607128372655             0   \n",
       "\n",
       "    verified_purchase  \n",
       "48               True  \n",
       "49               True  \n",
       "50               True  \n",
       "52               True  \n",
       "53               True  "
      ]
     },
     "execution_count": 217,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_set = ratings[~ratings.index.isin(test_set.index)]\n",
    "train_set.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### check if training set and test set are eligible for model training and testing\n",
    "* all users in train_set must have interacted at least 4 items in train_set (and have interacted at least 5 items in the whole dataset)  \n",
    "* all items in test_set must have been interacted (have appeared) in train_set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "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>count_asin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>9566.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>6.424420</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>3.006229</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>4.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>5.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>5.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>7.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>57.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        count_asin\n",
       "count  9566.000000\n",
       "mean      6.424420\n",
       "std       3.006229\n",
       "min       4.000000\n",
       "25%       5.000000\n",
       "50%       5.000000\n",
       "75%       7.000000\n",
       "max      57.000000"
      ]
     },
     "execution_count": 218,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_set.groupby('user_id').agg({'asin': 'nunique'}).reset_index().rename(columns={'asin': 'count_asin'}).describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0, 10)"
      ]
     },
     "execution_count": 222,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_set[~test_set.asin.isin(train_set.asin)].shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### filter and rename columns and save to files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "metadata": {},
   "outputs": [],
   "source": [
    "ratings = ratings[['user_id', 'asin', 'rating', 'timestamp']].rename(columns={'asin': 'item_id'})\n",
    "train_set = train_set[['user_id', 'asin', 'rating', 'timestamp']].rename(columns={'asin': 'item_id'})\n",
    "test_set = test_set[['user_id', 'asin', 'rating', 'timestamp']].rename(columns={'asin': 'item_id'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "metadata": {},
   "outputs": [],
   "source": [
    "ratings.to_csv('data/ratings.csv')\n",
    "train_set.to_csv('data/train_set.csv')\n",
    "test_set.to_csv('data/test_set.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Filter product metadata"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "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>main_category</th>\n",
       "      <th>title</th>\n",
       "      <th>average_rating</th>\n",
       "      <th>rating_number</th>\n",
       "      <th>features</th>\n",
       "      <th>description</th>\n",
       "      <th>price</th>\n",
       "      <th>images</th>\n",
       "      <th>videos</th>\n",
       "      <th>store</th>\n",
       "      <th>categories</th>\n",
       "      <th>details</th>\n",
       "      <th>parent_asin</th>\n",
       "      <th>bought_together</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>AMAZON FASHION</td>\n",
       "      <td>YUEDGE 5 Pairs Men's Moisture Control Cushione...</td>\n",
       "      <td>4.6</td>\n",
       "      <td>16</td>\n",
       "      <td>[]</td>\n",
       "      <td>[]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>[{'thumb': 'https://m.media-amazon.com/images/...</td>\n",
       "      <td>[]</td>\n",
       "      <td>GiveGift</td>\n",
       "      <td>[]</td>\n",
       "      <td>{'Package Dimensions': '10.31 x 8.5 x 1.73 inc...</td>\n",
       "      <td>B08BHN9PK5</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>AMAZON FASHION</td>\n",
       "      <td>DouBCQ Women's Palazzo Lounge Wide Leg Casual ...</td>\n",
       "      <td>4.1</td>\n",
       "      <td>7</td>\n",
       "      <td>[Drawstring closure, Machine Wash]</td>\n",
       "      <td>[]</td>\n",
       "      <td>NaN</td>\n",
       "      <td>[{'thumb': 'https://m.media-amazon.com/images/...</td>\n",
       "      <td>[]</td>\n",
       "      <td>DouBCQ</td>\n",
       "      <td>[]</td>\n",
       "      <td>{'Package Dimensions': '15 x 10.2 x 0.4 inches...</td>\n",
       "      <td>B08R39MRDW</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    main_category                                              title  \\\n",
       "0  AMAZON FASHION  YUEDGE 5 Pairs Men's Moisture Control Cushione...   \n",
       "1  AMAZON FASHION  DouBCQ Women's Palazzo Lounge Wide Leg Casual ...   \n",
       "\n",
       "   average_rating  rating_number                            features  \\\n",
       "0             4.6             16                                  []   \n",
       "1             4.1              7  [Drawstring closure, Machine Wash]   \n",
       "\n",
       "  description  price                                             images  \\\n",
       "0          []    NaN  [{'thumb': 'https://m.media-amazon.com/images/...   \n",
       "1          []    NaN  [{'thumb': 'https://m.media-amazon.com/images/...   \n",
       "\n",
       "  videos     store categories  \\\n",
       "0     []  GiveGift         []   \n",
       "1     []    DouBCQ         []   \n",
       "\n",
       "                                             details parent_asin  \\\n",
       "0  {'Package Dimensions': '10.31 x 8.5 x 1.73 inc...  B08BHN9PK5   \n",
       "1  {'Package Dimensions': '15 x 10.2 x 0.4 inches...  B08R39MRDW   \n",
       "\n",
       "  bought_together  \n",
       "0            None  \n",
       "1            None  "
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meta.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    826108.000000\n",
       "mean          0.181550\n",
       "std           1.374751\n",
       "min           0.000000\n",
       "25%           0.000000\n",
       "50%           0.000000\n",
       "75%           0.000000\n",
       "max         101.000000\n",
       "Name: description, dtype: float64"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meta.description.apply(lambda x: len(x)).describe()  ## description column not useful since too many are empty"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "metadata": {},
   "outputs": [],
   "source": [
    "product = meta[['parent_asin', 'title']].rename(columns={'parent_asin': 'item_id'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(52619, 2)"
      ]
     },
     "execution_count": 235,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "product = product[product.item_id.isin(ratings.item_id)]\n",
    "\n",
    "product.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>item_id</th>\n",
       "      <th>title</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>B07DXHY5ZX</td>\n",
       "      <td>PattyBoutik Women Crewneck Eyelet Cold Shoulde...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>B01AUYK33Y</td>\n",
       "      <td>Stainless Steel Bracelet Grooved Cuff Bangle f...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>B074C8GVHJ</td>\n",
       "      <td>ACEVOG Women Floral Printed Button Down V Neck...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>B077JWX1MX</td>\n",
       "      <td>ELEOSL Women Ladies Sleeveless Fit and Flare R...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>B08TWC3WKW</td>\n",
       "      <td>akimoom Scrunchie Watch Band for Galaxy 46mm/G...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       item_id                                              title\n",
       "9   B07DXHY5ZX  PattyBoutik Women Crewneck Eyelet Cold Shoulde...\n",
       "21  B01AUYK33Y  Stainless Steel Bracelet Grooved Cuff Bangle f...\n",
       "22  B074C8GVHJ  ACEVOG Women Floral Printed Button Down V Neck...\n",
       "26  B077JWX1MX  ELEOSL Women Ladies Sleeveless Fit and Flare R...\n",
       "39  B08TWC3WKW  akimoom Scrunchie Watch Band for Galaxy 46mm/G..."
      ]
     },
     "execution_count": 236,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "product.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [],
   "source": [
    "product.to_csv('data/product.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# EDA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 315,
   "metadata": {},
   "outputs": [],
   "source": [
    "ratings = pd.read_csv('data/ratings.csv')\n",
    "train_set = pd.read_csv('data/train_set.csv')\n",
    "test_set = pd.read_csv('data/test_set.csv')\n",
    "\n",
    "ratings.rename(columns={'Unnamed: 0': 'id'}, inplace=True)\n",
    "train_set.rename(columns={'Unnamed: 0': 'id'}, inplace=True)\n",
    "test_set.rename(columns={'Unnamed: 0': 'id'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 316,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((63229, 5), 9566, 52619)"
      ]
     },
     "execution_count": 316,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ratings.shape, ratings.user_id.nunique(), ratings.item_id.nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 317,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sparsity of ratings matrix: 0.00012561553329790666\n"
     ]
    }
   ],
   "source": [
    "print('sparsity of ratings matrix:', ratings.shape[0]/(ratings.user_id.nunique()*ratings.item_id.nunique()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 319,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>user_id</th>\n",
       "      <th>item_id</th>\n",
       "      <th>rating</th>\n",
       "      <th>timestamp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>48</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>B098NYMTYH</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1661808780730</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>49</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>B095CHR38Z</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1645062075013</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>50</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>B071HMN7K8</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1635309000822</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>52</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>B074XF4KG4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1611092662402</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>53</td>\n",
       "      <td>AFZUK3MTBIBEDQOPAK3OATUOUKLA</td>\n",
       "      <td>B08KXH42H9</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1607128372655</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   id                       user_id     item_id  rating      timestamp\n",
       "0  48  AFZUK3MTBIBEDQOPAK3OATUOUKLA  B098NYMTYH     2.0  1661808780730\n",
       "1  49  AFZUK3MTBIBEDQOPAK3OATUOUKLA  B095CHR38Z     2.0  1645062075013\n",
       "2  50  AFZUK3MTBIBEDQOPAK3OATUOUKLA  B071HMN7K8     5.0  1635309000822\n",
       "3  52  AFZUK3MTBIBEDQOPAK3OATUOUKLA  B074XF4KG4     1.0  1611092662402\n",
       "4  53  AFZUK3MTBIBEDQOPAK3OATUOUKLA  B08KXH42H9     4.0  1607128372655"
      ]
     },
     "execution_count": 319,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ratings.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 321,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8, 6))\n",
    "\n",
    "ratings['rating'].value_counts().sort_index().plot(kind='bar', figsize=(8, 6), edgecolor='k', alpha=0.7)\n",
    "plt.title('Distribution of Ratings')\n",
    "plt.xlabel('Rating')\n",
    "plt.ylabel('Frequency')\n",
    "plt.grid(True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model and Benchmark\n",
    "Google Colab: https://colab.research.google.com/drive/1CsePfKM1cga11LTNyFHwijytIAEtzwBi#scrollTo=hCBXS6Fa7jHg"
   ]
  }
 ],
 "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": "3.12.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}