examples/movielens.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import Required Packages and Set Options"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import sys\n",
    "import git\n",
    "\n",
    "import numpy as np\n",
    "import numba as nb\n",
    "import pandas as pd\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "import warnings\n",
    "# warnings.filterwarnings(\"ignore\", category=nb.NumbaPerformanceWarning)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "sns.set_style('whitegrid')\n",
    "plt.rcParams['axes.titlesize'] = 14\n",
    "plt.rcParams['axes.labelsize'] = 12\n",
    "plt.rcParams['xtick.labelsize'] = 12\n",
    "plt.rcParams['ytick.labelsize'] = 12\n",
    "# plt.rcParams"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Put the Main Package Library on the PYTHONPATH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['/Users/ericlundquist/Repos/rankfm',\n",
       " '/Users/ericlundquist/Repos/rankfm/rankfm']"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "git_repo = git.Repo('.', search_parent_directories=True)\n",
    "git_root = git_repo.git.rev_parse('--show-toplevel')\n",
    "cython_path = os.path.join(git_root, 'rankfm')\n",
    "\n",
    "sys.path[0] = git_root\n",
    "sys.path[1] = cython_path\n",
    "sys.path[:2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Re-Compile Cython Extension Module"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "re-generating C source with cythonize...\n",
      "Compiling rankfm/_rankfm.pyx because it changed.\n",
      "[1/1] Cythonizing rankfm/_rankfm.pyx\n",
      "running build_ext\n",
      "building 'rankfm._rankfm' extension\n",
      "gcc -Wno-unused-result -Wsign-compare -Wunreachable-code -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -I/Users/ericlundquist/anaconda3/include -arch x86_64 -I/Users/ericlundquist/anaconda3/include -arch x86_64 -I./rankfm -I/Users/ericlundquist/anaconda3/include/python3.7m -c rankfm/_rankfm.c -o build/temp.macosx-10.7-x86_64-3.7/rankfm/_rankfm.o -O2 -ffast-math -Wno-unused-function -Wno-uninitialized\n",
      "gcc -Wno-unused-result -Wsign-compare -Wunreachable-code -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -I/Users/ericlundquist/anaconda3/include -arch x86_64 -I/Users/ericlundquist/anaconda3/include -arch x86_64 -I./rankfm -I/Users/ericlundquist/anaconda3/include/python3.7m -c rankfm/mt19937ar/mt19937ar.c -o build/temp.macosx-10.7-x86_64-3.7/rankfm/mt19937ar/mt19937ar.o -O2 -ffast-math -Wno-unused-function -Wno-uninitialized\n",
      "gcc -bundle -undefined dynamic_lookup -L/Users/ericlundquist/anaconda3/lib -arch x86_64 -L/Users/ericlundquist/anaconda3/lib -arch x86_64 -arch x86_64 build/temp.macosx-10.7-x86_64-3.7/rankfm/_rankfm.o build/temp.macosx-10.7-x86_64-3.7/rankfm/mt19937ar/mt19937ar.o -o build/lib.macosx-10.7-x86_64-3.7/rankfm/_rankfm.cpython-37m-darwin.so\n",
      "copying build/lib.macosx-10.7-x86_64-3.7/rankfm/_rankfm.cpython-37m-darwin.so -> rankfm\n"
     ]
    }
   ],
   "source": [
    "!cd $git_root && python setup.py build_ext --inplace"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Dynamically Re-Load all Package Modules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "from rankfm.rankfm import RankFM\n",
    "from rankfm.evaluation import hit_rate, reciprocal_rank, discounted_cumulative_gain, precision, recall, diversity"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Set File Path Constants"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/Users/ericlundquist/Repos/rankfm\n",
      "/Users/ericlundquist/Repos/rankfm/data/examples\n"
     ]
    }
   ],
   "source": [
    "repo_root = \"/Users/ericlundquist/Repos/rankfm\"\n",
    "data_path = os.path.join(repo_root, \"data/examples\")\n",
    "print(\"\\n\".join([repo_root, data_path]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load Example Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Load Interactions Data\n",
    "* `interactions` must be a `pd.dataframe` or `np.ndarray` with columns: `[user_id, item_id]` when used for model fitting\n",
    "* the rows of `interactions` should represent observed or implied positive feedback for the user/item pair\n",
    "* examples of implicit feedback include product purchases, movie/article views, page views, clicks, etc.\n",
    "* the `user_id` and/or `item_id` identifiers can be arbitrary integers or even strings as long as they are unique identifiers for users/items\n",
    "* RankFM internally maps raw `user_id` and `item_id` values to zero-based index positions but maps all output back to original identifier values\n",
    "* it is recommended you perform any cross-validation or data splitting strategy requiring additional columns prior to preparing the interaction data for model fitting\n",
    "* in this example we'll assume all observed movie ratings count as positive implicit feedback regardless of numerical rating `[1-5]`\n",
    "* this most closely represents the intended data and modeling context for which RankFM was designed (implicit preferences recommendation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>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>1</td>\n",
       "      <td>1193</td>\n",
       "      <td>5</td>\n",
       "      <td>2000-12-31 22:12:40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>661</td>\n",
       "      <td>3</td>\n",
       "      <td>2000-12-31 22:35:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>914</td>\n",
       "      <td>3</td>\n",
       "      <td>2000-12-31 22:32:48</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>3408</td>\n",
       "      <td>4</td>\n",
       "      <td>2000-12-31 22:04:35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>2355</td>\n",
       "      <td>5</td>\n",
       "      <td>2001-01-06 23:38:11</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user_id  item_id  rating            timestamp\n",
       "0        1     1193       5  2000-12-31 22:12:40\n",
       "1        1      661       3  2000-12-31 22:35:09\n",
       "2        1      914       3  2000-12-31 22:32:48\n",
       "3        1     3408       4  2000-12-31 22:04:35\n",
       "4        1     2355       5  2001-01-06 23:38:11"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interactions = pd.read_csv(os.path.join(data_path, 'ML_1M_RATINGS.csv'))\n",
    "interactions.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Load User Features\n",
    "* `user_features` must be a `pd.dataframe` or `np.ndarray` with columns: `[user_id, uf_1, ... , uf_n]`\n",
    "* the first column must be the `user_id` and the set of `user_id` values must match the unique users found in `interactions`\n",
    "* the additional columns should be user attributes/features that can be coerced to `float32` data type (i.e. no strings)\n",
    "* care should be taken to scale any continuous features prior to model fitting for numerical stability - often dummy/tag features work best"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>gender__F</th>\n",
       "      <th>gender__M</th>\n",
       "      <th>agegrp__1</th>\n",
       "      <th>agegrp__18</th>\n",
       "      <th>agegrp__25</th>\n",
       "      <th>agegrp__35</th>\n",
       "      <th>agegrp__45</th>\n",
       "      <th>agegrp__50</th>\n",
       "      <th>agegrp__56</th>\n",
       "      <th>...</th>\n",
       "      <th>occupation__11</th>\n",
       "      <th>occupation__12</th>\n",
       "      <th>occupation__13</th>\n",
       "      <th>occupation__14</th>\n",
       "      <th>occupation__15</th>\n",
       "      <th>occupation__16</th>\n",
       "      <th>occupation__17</th>\n",
       "      <th>occupation__18</th>\n",
       "      <th>occupation__19</th>\n",
       "      <th>occupation__20</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 31 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   user_id  gender__F  gender__M  agegrp__1  agegrp__18  agegrp__25  \\\n",
       "0        1          1          0          1           0           0   \n",
       "1        2          0          1          0           0           0   \n",
       "2        3          0          1          0           0           1   \n",
       "3        4          0          1          0           0           0   \n",
       "4        5          0          1          0           0           1   \n",
       "\n",
       "   agegrp__35  agegrp__45  agegrp__50  agegrp__56  ...  occupation__11  \\\n",
       "0           0           0           0           0  ...               0   \n",
       "1           0           0           0           1  ...               0   \n",
       "2           0           0           0           0  ...               0   \n",
       "3           0           1           0           0  ...               0   \n",
       "4           0           0           0           0  ...               0   \n",
       "\n",
       "   occupation__12  occupation__13  occupation__14  occupation__15  \\\n",
       "0               0               0               0               0   \n",
       "1               0               0               0               0   \n",
       "2               0               0               0               1   \n",
       "3               0               0               0               0   \n",
       "4               0               0               0               0   \n",
       "\n",
       "   occupation__16  occupation__17  occupation__18  occupation__19  \\\n",
       "0               0               0               0               0   \n",
       "1               1               0               0               0   \n",
       "2               0               0               0               0   \n",
       "3               0               0               0               0   \n",
       "4               0               0               0               0   \n",
       "\n",
       "   occupation__20  \n",
       "0               0  \n",
       "1               0  \n",
       "2               0  \n",
       "3               0  \n",
       "4               1  \n",
       "\n",
       "[5 rows x 31 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_features = pd.read_csv(os.path.join(data_path, 'ML_1M_USERS.csv'))\n",
    "user_features.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Load Item Features\n",
    "* `item_features` must be a `pd.dataframe` or `np.ndarray` with columns: `[item_id, uf_1, ... , uf_n]`\n",
    "* the first column must be the `item_id` and the set of `item_id` values must match the unique items found in `interactions`\n",
    "* the additional columns should be item attributes/features that can be coerced to `float32` data type (i.e. no strings)\n",
    "* care should be taken to scale any continuous features prior to model fitting for numerical stability - often dummy/tag features work best"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>action</th>\n",
       "      <th>adventure</th>\n",
       "      <th>animation</th>\n",
       "      <th>childrens</th>\n",
       "      <th>comedy</th>\n",
       "      <th>crime</th>\n",
       "      <th>documentary</th>\n",
       "      <th>drama</th>\n",
       "      <th>fantasy</th>\n",
       "      <th>filmnoir</th>\n",
       "      <th>horror</th>\n",
       "      <th>musical</th>\n",
       "      <th>mystery</th>\n",
       "      <th>romance</th>\n",
       "      <th>scifi</th>\n",
       "      <th>thriller</th>\n",
       "      <th>war</th>\n",
       "      <th>western</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   item_id  action  adventure  animation  childrens  comedy  crime  \\\n",
       "0        1       0          0          1          1       1      0   \n",
       "1        2       0          1          0          1       0      0   \n",
       "2        3       0          0          0          0       1      0   \n",
       "3        4       0          0          0          0       1      0   \n",
       "4        5       0          0          0          0       1      0   \n",
       "\n",
       "   documentary  drama  fantasy  filmnoir  horror  musical  mystery  romance  \\\n",
       "0            0      0        0         0       0        0        0        0   \n",
       "1            0      0        1         0       0        0        0        0   \n",
       "2            0      0        0         0       0        0        0        1   \n",
       "3            0      1        0         0       0        0        0        0   \n",
       "4            0      0        0         0       0        0        0        0   \n",
       "\n",
       "   scifi  thriller  war  western  \n",
       "0      0         0    0        0  \n",
       "1      0         0    0        0  \n",
       "2      0         0    0        0  \n",
       "3      0         0    0        0  \n",
       "4      0         0    0        0  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "item_features = pd.read_csv(os.path.join(data_path, 'ML_1M_ITEMS.csv'))\n",
    "item_features.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Load Item Names\n",
    "* these won't be used for model fitting, but are important when reviewing the output for face validity\n",
    "* also analyzing recommended/similar movies isn't any fun if you can't see the titles!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>item_id</th>\n",
       "      <th>title</th>\n",
       "      <th>genres</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>Toy Story (1995)</td>\n",
       "      <td>Animation|Children's|Comedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>Jumanji (1995)</td>\n",
       "      <td>Adventure|Children's|Fantasy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>Grumpier Old Men (1995)</td>\n",
       "      <td>Comedy|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>Waiting to Exhale (1995)</td>\n",
       "      <td>Comedy|Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>Father of the Bride Part II (1995)</td>\n",
       "      <td>Comedy</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   item_id                               title                        genres\n",
       "0        1                    Toy Story (1995)   Animation|Children's|Comedy\n",
       "1        2                      Jumanji (1995)  Adventure|Children's|Fantasy\n",
       "2        3             Grumpier Old Men (1995)                Comedy|Romance\n",
       "3        4            Waiting to Exhale (1995)                  Comedy|Drama\n",
       "4        5  Father of the Bride Part II (1995)                        Comedy"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "item_names = pd.read_csv(os.path.join(data_path, 'ML_1M_ITEM_NAMES.csv'))\n",
    "item_names.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Check Matrix/Vector Dimensions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "interactions shape: (1000209, 4)\n",
      "interactions unique users: 6040\n",
      "interactions unique items: 3706\n",
      "user features users: 6040\n",
      "item features items: 3706\n"
     ]
    }
   ],
   "source": [
    "unique_users = interactions.user_id.nunique()\n",
    "unique_items = interactions.item_id.nunique()\n",
    "\n",
    "print(\"interactions shape: {}\".format(interactions.shape))\n",
    "print(\"interactions unique users: {}\".format(interactions.user_id.nunique()))\n",
    "print(\"interactions unique items: {}\".format(interactions.item_id.nunique()))\n",
    "\n",
    "print(\"user features users:\", interactions.user_id.nunique())\n",
    "print(\"item features items:\", interactions.item_id.nunique())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Evaluate Interaction Matrix Sparsity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "interaction matrix sparsity: 95.5%\n"
     ]
    }
   ],
   "source": [
    "sparsity = 1 - (len(interactions) / (unique_users * unique_items))\n",
    "print(\"interaction matrix sparsity: {}%\".format(round(100 * sparsity, 1)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Split Interactions Data into (Train, Valid) for Model Evaluation\n",
    "* we'll randomly split off 25% of the observed user/item interactions for model validation\n",
    "* take care to note any users/items in the validation set not appearing in the training set\n",
    "* for model fitting we'll have to unify the set of users/items across `interactions`, `user_features`, `item_features`\n",
    "* for scoring/recommendation RankFM has the `cold_start` option to either drop unseen users/items or set model output to np.nan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.random.seed(1492)\n",
    "interactions['random'] = np.random.random(size=len(interactions))\n",
    "test_pct = 0.25"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train shape: (749724, 2)\n",
      "valid shape: (250485, 2)\n",
      "train users: 6040\n",
      "valid users: 6038\n",
      "cold-start users: set()\n",
      "train items: 3666\n",
      "valid items: 3531\n",
      "cold-start items: {3202, 2563, 644, 3460, 3209, 3722, 139, 396, 651, 1555, 1434, 1915, 1830, 2217, 2218, 687, 1842, 2742, 1851, 2619, 576, 3321, 3277, 3323, 3280, 601, 3290, 3164, 989, 865, 226, 3172, 1386, 878, 3187, 758, 3065, 763, 2556, 127}\n"
     ]
    }
   ],
   "source": [
    "train_mask = interactions['random'] <  (1 - test_pct)\n",
    "valid_mask = interactions['random'] >= (1 - test_pct)\n",
    "\n",
    "interactions_train = interactions[train_mask][['user_id', 'item_id']]\n",
    "interactions_valid = interactions[valid_mask][['user_id', 'item_id']]\n",
    "\n",
    "train_users = np.sort(interactions_train.user_id.unique())\n",
    "valid_users = np.sort(interactions_valid.user_id.unique())\n",
    "cold_start_users = set(valid_users) - set(train_users)\n",
    "\n",
    "train_items = np.sort(interactions_train.item_id.unique())\n",
    "valid_items = np.sort(interactions_valid.item_id.unique())\n",
    "cold_start_items = set(valid_items) - set(train_items)\n",
    "\n",
    "print(\"train shape: {}\".format(interactions_train.shape))\n",
    "print(\"valid shape: {}\".format(interactions_valid.shape))\n",
    "\n",
    "print(\"train users: {}\".format(len(train_users)))\n",
    "print(\"valid users: {}\".format(len(valid_users)))\n",
    "print(\"cold-start users: {}\".format(cold_start_users))\n",
    "\n",
    "print(\"train items: {}\".format(len(train_items)))\n",
    "print(\"valid items: {}\".format(len(valid_items)))\n",
    "print(\"cold-start items: {}\".format(cold_start_items))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((6040, 31), (3666, 19))"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_features = user_features[user_features.user_id.isin(train_users)]\n",
    "item_features = item_features[item_features.item_id.isin(train_items)]\n",
    "user_features.shape, item_features.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Fit the Model on the Training Data and Evaluate Out-of-Sample Performance Metrics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Initialize the Model Object"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<rankfm.rankfm.RankFM at 0x10c753ac8>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = RankFM(factors=20, loss='warp', max_samples=20, alpha=0.01, sigma=0.1, learning_rate=0.10, learning_schedule='invscaling')\n",
    "model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Fit the Model on the Training Data\n",
    "* each epoch represents one full pass through all observed user/item interactions in the training data\n",
    "* the `verbose` flag will print the current training epoch as well as the epoch's penalized log-likelihood value\n",
    "* the training algorithm attempts to maximize the model's log likelihood which represents user preferences for observed items over unobserved items\n",
    "* the log likelihood isn't 1:1 correlated with our recommendation/ranking evaluation metrics but serves as a decent proxy for training progress "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "training epoch: 0\n",
      "log likelihood: -434433.75\n",
      "\n",
      "training epoch: 1\n",
      "log likelihood: -420144.03125\n",
      "\n",
      "training epoch: 2\n",
      "log likelihood: -410680.4375\n",
      "\n",
      "training epoch: 3\n",
      "log likelihood: -405669.09375\n",
      "\n",
      "training epoch: 4\n",
      "log likelihood: -402369.28125\n",
      "\n",
      "training epoch: 5\n",
      "log likelihood: -400163.34375\n",
      "\n",
      "training epoch: 6\n",
      "log likelihood: -398232.21875\n",
      "\n",
      "training epoch: 7\n",
      "log likelihood: -396977.09375\n",
      "\n",
      "training epoch: 8\n",
      "log likelihood: -395550.78125\n",
      "\n",
      "training epoch: 9\n",
      "log likelihood: -394704.4375\n",
      "\n",
      "training epoch: 10\n",
      "log likelihood: -393696.75\n",
      "\n",
      "training epoch: 11\n",
      "log likelihood: -393200.71875\n",
      "\n",
      "training epoch: 12\n",
      "log likelihood: -392413.6875\n",
      "\n",
      "training epoch: 13\n",
      "log likelihood: -392190.46875\n",
      "\n",
      "training epoch: 14\n",
      "log likelihood: -391532.78125\n",
      "\n",
      "training epoch: 15\n",
      "log likelihood: -391579.0625\n",
      "\n",
      "training epoch: 16\n",
      "log likelihood: -391001.6875\n",
      "\n",
      "training epoch: 17\n",
      "log likelihood: -390895.0625\n",
      "\n",
      "training epoch: 18\n",
      "log likelihood: -390115.15625\n",
      "\n",
      "training epoch: 19\n",
      "log likelihood: -390569.1875\n",
      "CPU times: user 34.1 s, sys: 217 ms, total: 34.3 s\n",
      "Wall time: 29.7 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "model.fit(interactions_train, epochs=20, verbose=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Generate Model Scores for Validation Interactions\n",
    "* the `predict()` method generates real-valued model scores for user/item pairs\n",
    "* specify scoring interactions or user/item pairs the same way as training: `[user_id, item_id]`\n",
    "* the scores generated are those produced by the underlying second-degree factorization machine model\n",
    "* the `cold_start` option allows you to either generate missing values for unseen users/items or drop them from scoring output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(250485,)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "count    250437.000000\n",
       "mean          0.734153\n",
       "std           0.917317\n",
       "min          -3.192067\n",
       "25%           0.095812\n",
       "50%           0.725559\n",
       "75%           1.373034\n",
       "max           4.098247\n",
       "dtype: float64"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "valid_scores = model.predict(interactions_valid, cold_start='nan') \n",
    "print(valid_scores.shape)\n",
    "pd.Series(valid_scores).describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Generate TopN Recommendations for Validation Users\n",
    "* instead of having to generate predictions for all user/item pairs the `recommend()` method lets you generate the topN recommended items for each user\n",
    "* the output is a pd.dataframe where the index values are `user_id` and the columns are recommended items in descending order (first column is best item)\n",
    "* additionally there's an option `filter_previous` to exclude any items already observed during training for the user from being recommended\n",
    "* as with `predict()` there's a `cold_start` option to either fill with missing or drop recommendations for any user not found in training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>364</td>\n",
       "      <td>34</td>\n",
       "      <td>318</td>\n",
       "      <td>2858</td>\n",
       "      <td>1704</td>\n",
       "      <td>2081</td>\n",
       "      <td>593</td>\n",
       "      <td>912</td>\n",
       "      <td>2396</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>527</td>\n",
       "      <td>1610</td>\n",
       "      <td>260</td>\n",
       "      <td>3418</td>\n",
       "      <td>733</td>\n",
       "      <td>590</td>\n",
       "      <td>150</td>\n",
       "      <td>377</td>\n",
       "      <td>1704</td>\n",
       "      <td>296</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1197</td>\n",
       "      <td>1</td>\n",
       "      <td>2804</td>\n",
       "      <td>110</td>\n",
       "      <td>2791</td>\n",
       "      <td>2918</td>\n",
       "      <td>2028</td>\n",
       "      <td>2000</td>\n",
       "      <td>1387</td>\n",
       "      <td>2571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2571</td>\n",
       "      <td>1387</td>\n",
       "      <td>1210</td>\n",
       "      <td>1198</td>\n",
       "      <td>1196</td>\n",
       "      <td>858</td>\n",
       "      <td>924</td>\n",
       "      <td>2640</td>\n",
       "      <td>589</td>\n",
       "      <td>1221</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2396</td>\n",
       "      <td>2336</td>\n",
       "      <td>2692</td>\n",
       "      <td>2599</td>\n",
       "      <td>223</td>\n",
       "      <td>3160</td>\n",
       "      <td>2318</td>\n",
       "      <td>2712</td>\n",
       "      <td>1041</td>\n",
       "      <td>2912</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      0     1     2     3     4     5     6     7     8     9\n",
       "1     1   364    34   318  2858  1704  2081   593   912  2396\n",
       "2   527  1610   260  3418   733   590   150   377  1704   296\n",
       "3  1197     1  2804   110  2791  2918  2028  2000  1387  2571\n",
       "4  2571  1387  1210  1198  1196   858   924  2640   589  1221\n",
       "5  2396  2336  2692  2599   223  3160  2318  2712  1041  2912"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "valid_recommendations = model.recommend(valid_users, n_items=10, filter_previous=True, cold_start='nan')\n",
    "valid_recommendations.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Evaluate Model Performance on the Validation Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Specify Number of Recommended Items\n",
    "* here we'll look at model performance in terms of validation metrics wrt the top 10 recommended items for each user\n",
    "* it's good to compare validation metrics against the best unpersonalized recommendations: overall item popularity rankings from the training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "k = 10"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Generate Pure-Popularity Baselines"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "item_id\n",
       "2858    2545\n",
       "260     2254\n",
       "1196    2230\n",
       "1210    2178\n",
       "589     2002\n",
       "2028    1980\n",
       "480     1974\n",
       "1270    1939\n",
       "2571    1921\n",
       "608     1910\n",
       "Name: user_id, dtype: int64"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "most_popular = interactions_train.groupby('item_id')['user_id'].count().sort_values(ascending=False)[:k]\n",
    "most_popular"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "number of test users: 6038\n",
      "baseline hit rate: 0.635\n",
      "baseline precision: 0.116\n",
      "baseline recall: 0.046\n"
     ]
    }
   ],
   "source": [
    "test_user_items = interactions_valid.groupby('user_id')['item_id'].apply(set).to_dict()\n",
    "test_user_items = {key: val for key, val in test_user_items.items() if key in set(train_users)}\n",
    "\n",
    "base_hrt = np.mean([int(len(set(most_popular.index) & set(val)) > 0)                       for key, val in test_user_items.items()])\n",
    "base_pre = np.mean([len(set(most_popular.index) & set(val)) / len(set(most_popular.index)) for key, val in test_user_items.items()])\n",
    "base_rec = np.mean([len(set(most_popular.index) & set(val)) / len(set(val))                for key, val in test_user_items.items()])\n",
    "\n",
    "print(\"number of test users: {}\".format(len(test_user_items)))\n",
    "print(\"baseline hit rate: {:.3f}\".format(base_hrt))\n",
    "print(\"baseline precision: {:.3f}\".format(base_pre))\n",
    "print(\"baseline recall: {:.3f}\".format(base_rec))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Generate Model Validation Metrics\n",
    "* the RankFM package comes with a number of utility functions to evaluate model performance\n",
    "* these functions cover common validation metrics for recommendation/ranking models with implicit feedback data\n",
    "* all evaluation functions have the same interface and require a previously fit RankFM model\n",
    "    * `hit_rate`: the proportion of users with any relevant recommended item\n",
    "    * `reciprocal_rank`: the inverse rank of the first relevant recommended item (or zero if no matches)\n",
    "    * `discounted_cumulative_gain`: the sum of the inverse log-scaled ranks for relevant recommended items\n",
    "    * `precision`: the number of relevant recommended items / number of recommended items\n",
    "    * `recall`: the number of relevant recommended items / number of relevant items"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 23 s, sys: 118 ms, total: 23.1 s\n",
      "Wall time: 23.1 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "model_hit_rate = hit_rate(model, interactions_valid, k=k)\n",
    "model_reciprocal_rank = reciprocal_rank(model, interactions_valid, k=k)\n",
    "model_dcg = discounted_cumulative_gain(model, interactions_valid, k=k)\n",
    "model_precision = precision(model, interactions_valid, k=k)\n",
    "model_recall = recall(model, interactions_valid, k=k)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "hit_rate: 0.798\n",
      "reciprocal_rank: 0.341\n",
      "dcg: 0.733\n",
      "precision: 0.159\n",
      "recall: 0.077\n"
     ]
    }
   ],
   "source": [
    "print(\"hit_rate: {:.3f}\".format(model_hit_rate))\n",
    "print(\"reciprocal_rank: {:.3f}\".format(model_reciprocal_rank))\n",
    "print(\"dcg: {:.3f}\".format(model_dcg, 3))\n",
    "print(\"precision: {:.3f}\".format(model_precision))\n",
    "print(\"recall: {:.3f}\".format(model_recall))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Assess Recommendation Diversity\n",
    "* in addition to relevance, it's often important to assess the diversity of model recommendations\n",
    "* in other words, how many items are getting recommended overall, what's the percentage of users getting recommended the top items, etc.\n",
    "* this is an important component of personalization - high diversity will help prevent the harmful rich-get-richer feedback loop over time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>item_id</th>\n",
       "      <th>cnt_users</th>\n",
       "      <th>pct_users</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2858</td>\n",
       "      <td>2014</td>\n",
       "      <td>0.333554</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1196</td>\n",
       "      <td>1498</td>\n",
       "      <td>0.248095</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>296</td>\n",
       "      <td>1327</td>\n",
       "      <td>0.219775</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>260</td>\n",
       "      <td>1318</td>\n",
       "      <td>0.218284</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2571</td>\n",
       "      <td>1257</td>\n",
       "      <td>0.208182</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2762</td>\n",
       "      <td>1226</td>\n",
       "      <td>0.203047</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>608</td>\n",
       "      <td>1198</td>\n",
       "      <td>0.198410</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1198</td>\n",
       "      <td>1190</td>\n",
       "      <td>0.197085</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2997</td>\n",
       "      <td>1187</td>\n",
       "      <td>0.196588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>589</td>\n",
       "      <td>1176</td>\n",
       "      <td>0.194766</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   item_id  cnt_users  pct_users\n",
       "0     2858       2014   0.333554\n",
       "1     1196       1498   0.248095\n",
       "2      296       1327   0.219775\n",
       "3      260       1318   0.218284\n",
       "4     2571       1257   0.208182\n",
       "5     2762       1226   0.203047\n",
       "6      608       1198   0.198410\n",
       "7     1198       1190   0.197085\n",
       "8     2997       1187   0.196588\n",
       "9      589       1176   0.194766"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "recommendation_diversity = diversity(model, interactions_valid, k=k)\n",
    "recommendation_diversity.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead 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>cnt_users</th>\n",
       "      <th>pct_users</th>\n",
       "      <th>title</th>\n",
       "      <th>genres</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2858</td>\n",
       "      <td>2014</td>\n",
       "      <td>0.333554</td>\n",
       "      <td>American Beauty (1999)</td>\n",
       "      <td>Comedy|Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1196</td>\n",
       "      <td>1498</td>\n",
       "      <td>0.248095</td>\n",
       "      <td>Star Wars: Episode V - The Empire Strikes Back...</td>\n",
       "      <td>Action|Adventure|Drama|Sci-Fi|War</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>296</td>\n",
       "      <td>1327</td>\n",
       "      <td>0.219775</td>\n",
       "      <td>Pulp Fiction (1994)</td>\n",
       "      <td>Crime|Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>260</td>\n",
       "      <td>1318</td>\n",
       "      <td>0.218284</td>\n",
       "      <td>Star Wars: Episode IV - A New Hope (1977)</td>\n",
       "      <td>Action|Adventure|Fantasy|Sci-Fi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2571</td>\n",
       "      <td>1257</td>\n",
       "      <td>0.208182</td>\n",
       "      <td>Matrix, The (1999)</td>\n",
       "      <td>Action|Sci-Fi|Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2762</td>\n",
       "      <td>1226</td>\n",
       "      <td>0.203047</td>\n",
       "      <td>Sixth Sense, The (1999)</td>\n",
       "      <td>Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>608</td>\n",
       "      <td>1198</td>\n",
       "      <td>0.198410</td>\n",
       "      <td>Fargo (1996)</td>\n",
       "      <td>Crime|Drama|Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1198</td>\n",
       "      <td>1190</td>\n",
       "      <td>0.197085</td>\n",
       "      <td>Raiders of the Lost Ark (1981)</td>\n",
       "      <td>Action|Adventure</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2997</td>\n",
       "      <td>1187</td>\n",
       "      <td>0.196588</td>\n",
       "      <td>Being John Malkovich (1999)</td>\n",
       "      <td>Comedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>589</td>\n",
       "      <td>1176</td>\n",
       "      <td>0.194766</td>\n",
       "      <td>Terminator 2: Judgment Day (1991)</td>\n",
       "      <td>Action|Sci-Fi|Thriller</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   item_id  cnt_users  pct_users  \\\n",
       "0     2858       2014   0.333554   \n",
       "1     1196       1498   0.248095   \n",
       "2      296       1327   0.219775   \n",
       "3      260       1318   0.218284   \n",
       "4     2571       1257   0.208182   \n",
       "5     2762       1226   0.203047   \n",
       "6      608       1198   0.198410   \n",
       "7     1198       1190   0.197085   \n",
       "8     2997       1187   0.196588   \n",
       "9      589       1176   0.194766   \n",
       "\n",
       "                                               title  \\\n",
       "0                             American Beauty (1999)   \n",
       "1  Star Wars: Episode V - The Empire Strikes Back...   \n",
       "2                                Pulp Fiction (1994)   \n",
       "3          Star Wars: Episode IV - A New Hope (1977)   \n",
       "4                                 Matrix, The (1999)   \n",
       "5                            Sixth Sense, The (1999)   \n",
       "6                                       Fargo (1996)   \n",
       "7                     Raiders of the Lost Ark (1981)   \n",
       "8                        Being John Malkovich (1999)   \n",
       "9                  Terminator 2: Judgment Day (1991)   \n",
       "\n",
       "                              genres  \n",
       "0                       Comedy|Drama  \n",
       "1  Action|Adventure|Drama|Sci-Fi|War  \n",
       "2                        Crime|Drama  \n",
       "3    Action|Adventure|Fantasy|Sci-Fi  \n",
       "4             Action|Sci-Fi|Thriller  \n",
       "5                           Thriller  \n",
       "6               Crime|Drama|Thriller  \n",
       "7                   Action|Adventure  \n",
       "8                             Comedy  \n",
       "9             Action|Sci-Fi|Thriller  "
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_items = pd.merge(item_names, recommendation_diversity, on='item_id', how='inner')\n",
    "top_items = top_items.set_index('item_id').loc[recommendation_diversity.item_id].reset_index()\n",
    "top_items = top_items[['item_id', 'cnt_users', 'pct_users', 'title', 'genres']]\n",
    "top_items.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "percentage of items recommended to at least one user: 0.204\n"
     ]
    }
   ],
   "source": [
    "coverage = np.mean(recommendation_diversity['cnt_users'] > 0)\n",
    "print(\"percentage of items recommended to at least one user: {:.3f}\".format(coverage))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "entropy value of recommended items: 41.571\n"
     ]
    }
   ],
   "source": [
    "nonzero_users = recommendation_diversity[recommendation_diversity.cnt_users > 0]\n",
    "entropy = -np.sum(nonzero_users['pct_users'] * np.log2(nonzero_users['pct_users']))\n",
    "print(\"entropy value of recommended items: {:.3f}\".format(entropy))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "N = 50\n",
    "fig, axes = plt.subplots(1, 1, figsize=[16, 4])\n",
    "\n",
    "topN = recommendation_diversity.iloc[:N, :]\n",
    "axes.bar(topN.index.values + 1, topN.pct_users, width=1, edgecolor='black', alpha=0.75)\n",
    "axes.set(xlabel='Item Rank', ylabel='Percentage of Users', title='Percentage of Users Recommended by Item Rank')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Calculate the Most Similar Items to a Given Item\n",
    "* the model's representation of users/items in latent factor space allows for the calculation of item-item similarity with simple dot products\n",
    "* in many contexts it's useful to serve similar items either for user acceptance or for contextual recommendations \n",
    "* one example might be recommending the user a list of movies to watch next after he/she has just finished watching a given movie"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "random user: 467\n"
     ]
    }
   ],
   "source": [
    "random_user = np.random.choice(interactions_valid.user_id.unique())\n",
    "print(\"random user: {}\".format(random_user))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>title</th>\n",
       "      <th>genres</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>item_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>597</th>\n",
       "      <td>Pretty Woman (1990)</td>\n",
       "      <td>Comedy|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>539</th>\n",
       "      <td>Sleepless in Seattle (1993)</td>\n",
       "      <td>Comedy|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>587</th>\n",
       "      <td>Ghost (1990)</td>\n",
       "      <td>Comedy|Romance|Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1569</th>\n",
       "      <td>My Best Friend's Wedding (1997)</td>\n",
       "      <td>Comedy|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>357</th>\n",
       "      <td>Four Weddings and a Funeral (1994)</td>\n",
       "      <td>Comedy|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2291</th>\n",
       "      <td>Edward Scissorhands (1990)</td>\n",
       "      <td>Drama|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1968</th>\n",
       "      <td>Breakfast Club, The (1985)</td>\n",
       "      <td>Comedy|Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2396</th>\n",
       "      <td>Shakespeare in Love (1998)</td>\n",
       "      <td>Comedy|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1265</th>\n",
       "      <td>Groundhog Day (1993)</td>\n",
       "      <td>Comedy|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2762</th>\n",
       "      <td>Sixth Sense, The (1999)</td>\n",
       "      <td>Thriller</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                      title                   genres\n",
       "item_id                                                             \n",
       "597                     Pretty Woman (1990)           Comedy|Romance\n",
       "539             Sleepless in Seattle (1993)           Comedy|Romance\n",
       "587                            Ghost (1990)  Comedy|Romance|Thriller\n",
       "1569        My Best Friend's Wedding (1997)           Comedy|Romance\n",
       "357      Four Weddings and a Funeral (1994)           Comedy|Romance\n",
       "2291             Edward Scissorhands (1990)            Drama|Romance\n",
       "1968             Breakfast Club, The (1985)             Comedy|Drama\n",
       "2396             Shakespeare in Love (1998)           Comedy|Romance\n",
       "1265                   Groundhog Day (1993)           Comedy|Romance\n",
       "2762                Sixth Sense, The (1999)                 Thriller"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "random_user_recs = valid_recommendations.loc[random_user]\n",
    "random_user_recs = item_names[item_names.item_id.isin(random_user_recs)].set_index('item_id').loc[random_user_recs]\n",
    "random_user_recs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Look at Similar Movies for a Few of the User's Recommended Movies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "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>title</th>\n",
       "      <th>genres</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>item_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>594</th>\n",
       "      <td>Snow White and the Seven Dwarfs (1937)</td>\n",
       "      <td>Animation|Children's|Musical</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>899</th>\n",
       "      <td>Singin' in the Rain (1952)</td>\n",
       "      <td>Musical|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>914</th>\n",
       "      <td>My Fair Lady (1964)</td>\n",
       "      <td>Musical|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>904</th>\n",
       "      <td>Rear Window (1954)</td>\n",
       "      <td>Mystery|Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1282</th>\n",
       "      <td>Fantasia (1940)</td>\n",
       "      <td>Animation|Children's|Musical</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1029</th>\n",
       "      <td>Dumbo (1941)</td>\n",
       "      <td>Animation|Children's|Musical</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1207</th>\n",
       "      <td>To Kill a Mockingbird (1962)</td>\n",
       "      <td>Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1247</th>\n",
       "      <td>Graduate, The (1967)</td>\n",
       "      <td>Drama|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1947</th>\n",
       "      <td>West Side Story (1961)</td>\n",
       "      <td>Musical|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3549</th>\n",
       "      <td>Guys and Dolls (1955)</td>\n",
       "      <td>Musical</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                          title                        genres\n",
       "item_id                                                                      \n",
       "594      Snow White and the Seven Dwarfs (1937)  Animation|Children's|Musical\n",
       "899                  Singin' in the Rain (1952)               Musical|Romance\n",
       "914                         My Fair Lady (1964)               Musical|Romance\n",
       "904                          Rear Window (1954)              Mystery|Thriller\n",
       "1282                            Fantasia (1940)  Animation|Children's|Musical\n",
       "1029                               Dumbo (1941)  Animation|Children's|Musical\n",
       "1207               To Kill a Mockingbird (1962)                         Drama\n",
       "1247                       Graduate, The (1967)                 Drama|Romance\n",
       "1947                     West Side Story (1961)               Musical|Romance\n",
       "3549                      Guys and Dolls (1955)                       Musical"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "most_similar_items = model.similar_items(919)\n",
    "most_similar_items = item_names.set_index('item_id').loc[most_similar_items]\n",
    "most_similar_items"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "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>title</th>\n",
       "      <th>genres</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>item_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>110</th>\n",
       "      <td>Braveheart (1995)</td>\n",
       "      <td>Action|Drama|War</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>318</th>\n",
       "      <td>Shawshank Redemption, The (1994)</td>\n",
       "      <td>Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2890</th>\n",
       "      <td>Three Kings (1999)</td>\n",
       "      <td>Drama|War</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2571</th>\n",
       "      <td>Matrix, The (1999)</td>\n",
       "      <td>Action|Sci-Fi|Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>296</th>\n",
       "      <td>Pulp Fiction (1994)</td>\n",
       "      <td>Crime|Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>527</th>\n",
       "      <td>Schindler's List (1993)</td>\n",
       "      <td>Drama|War</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1213</th>\n",
       "      <td>GoodFellas (1990)</td>\n",
       "      <td>Crime|Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>593</th>\n",
       "      <td>Silence of the Lambs, The (1991)</td>\n",
       "      <td>Drama|Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1704</th>\n",
       "      <td>Good Will Hunting (1997)</td>\n",
       "      <td>Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1036</th>\n",
       "      <td>Die Hard (1988)</td>\n",
       "      <td>Action|Thriller</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                    title                  genres\n",
       "item_id                                                          \n",
       "110                     Braveheart (1995)        Action|Drama|War\n",
       "318      Shawshank Redemption, The (1994)                   Drama\n",
       "2890                   Three Kings (1999)               Drama|War\n",
       "2571                   Matrix, The (1999)  Action|Sci-Fi|Thriller\n",
       "296                   Pulp Fiction (1994)             Crime|Drama\n",
       "527               Schindler's List (1993)               Drama|War\n",
       "1213                    GoodFellas (1990)             Crime|Drama\n",
       "593      Silence of the Lambs, The (1991)          Drama|Thriller\n",
       "1704             Good Will Hunting (1997)                   Drama\n",
       "1036                      Die Hard (1988)         Action|Thriller"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "most_similar_items = model.similar_items(2028)\n",
    "most_similar_items = item_names.set_index('item_id').loc[most_similar_items]\n",
    "most_similar_items"
   ]
  },
  {
   "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": 4
}