graph.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this notebook I try a different approach for the recommendation, based on performing random walks on the graph that links users to the group in which they have been, with the number of messages sent as weight of the connection."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import networkx as nx\n",
    "import csrgraph as cg\n",
    "import numpy as np\n",
    "from tqdm.notebook import tqdm\n",
    "from joblib import Parallel, delayed\n",
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>group_id</th>\n",
       "      <th>messages_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>190</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001568744489</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1802</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001461601993</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001405226631</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2203</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001342690802</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3196</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001357462160</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        user_id       group_id  messages_count\n",
       "190   350313104 -1001568744489               3\n",
       "1802  350313104 -1001461601993               2\n",
       "2018  350313104 -1001405226631               3\n",
       "2203  350313104 -1001342690802               7\n",
       "3196  350313104 -1001357462160               9"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "group_membership = pd.read_pickle(\"df_membership.pkl\")\n",
    "users = pd.read_pickle(\"users.pkl\")\n",
    "groups = pd.read_pickle(\"groups.pickle\")\n",
    "group_membership = group_membership.loc[:,[\"user_id\", \"group_id\",\"messages_count\"]]\n",
    "group_membership[group_membership.user_id==350313104].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "G = nx.Graph()\n",
    "for _,row in group_membership.iterrows():\n",
    "    G.add_edge(row.user_id, row.group_id, weight=row.messages_count)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For each node (representing a user or a group) I perform 50 random walks of length 6."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[    1292286374, -1001568744489,      903461122, -1001568744489,\n",
       "             806726832, -1001568744489],\n",
       "       [-1001568744489,       66579321, -1001489401579,      187927651,\n",
       "        -1001254729041,      439203500]])"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "G = cg.csrgraph(G)\n",
    "node_names = G.names\n",
    "random_walks = G.random_walks(walklen=6, epochs=50, return_weight=1, neighbor_weight=1)\n",
    "labeled_walks = np.vectorize(lambda x : node_names[x])(random_walks)\n",
    "labeled_walks[:2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then I save the start and end of each walk with the number of times a walk from the node ended there."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>start</th>\n",
       "      <th>ends</th>\n",
       "      <th>visits</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-1001787166958</td>\n",
       "      <td>64701764</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-1001787166958</td>\n",
       "      <td>113374506</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-1001787166958</td>\n",
       "      <td>136953086</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-1001787166958</td>\n",
       "      <td>144897765</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-1001787166958</td>\n",
       "      <td>278035059</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           start       ends  visits\n",
       "0 -1001787166958   64701764       1\n",
       "1 -1001787166958  113374506       1\n",
       "2 -1001787166958  136953086       1\n",
       "3 -1001787166958  144897765       1\n",
       "4 -1001787166958  278035059       2"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "start_end_points = list()\n",
    "for walk in labeled_walks:\n",
    "    start_end_points.append((walk[0], walk[len(walk)-1]))\n",
    "\n",
    "walk_results = pd.DataFrame(start_end_points)\\\n",
    "            .rename({0:\"start\", 1:\"ends\"}, axis=1)\\\n",
    "            .reset_index()\\\n",
    "            .groupby([\"start\",\"ends\"])\\\n",
    "            .count()\\\n",
    "            .reset_index()\\\n",
    "            .rename({\"index\":\"visits\"}, axis=1)\n",
    "walk_results.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user_id</th>\n",
       "      <th>group_id</th>\n",
       "      <th>visits</th>\n",
       "      <th>first_name</th>\n",
       "      <th>last_name</th>\n",
       "      <th>username</th>\n",
       "      <th>title</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>12116</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001456212600</td>\n",
       "      <td>4</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Chat | Studenti UniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19169</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001396181733</td>\n",
       "      <td>4</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Algoritmi e strutture dati - Informatica</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49409</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001388941161</td>\n",
       "      <td>3</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Matematica del continuo - Informatica</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23258</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001379726039</td>\n",
       "      <td>3</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Ingegneria del software [PSS + PMD] (Club del ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>161472</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001569149223</td>\n",
       "      <td>2</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Lingue e letterature straniere | StudentiUniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>157771</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001699979466</td>\n",
       "      <td>2</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>International students | StudentiUniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>68973</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001157628331</td>\n",
       "      <td>2</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Programmazione 1 - Informatica, Informatica Mu...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38242</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001167949464</td>\n",
       "      <td>2</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Alloggi - StudentiUniMi</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         user_id       group_id  visits first_name last_name    username  \\\n",
       "12116   26170256 -1001456212600       4      Marco     Aceti  acetimarco   \n",
       "19169   26170256 -1001396181733       4      Marco     Aceti  acetimarco   \n",
       "49409   26170256 -1001388941161       3      Marco     Aceti  acetimarco   \n",
       "23258   26170256 -1001379726039       3      Marco     Aceti  acetimarco   \n",
       "161472  26170256 -1001569149223       2      Marco     Aceti  acetimarco   \n",
       "157771  26170256 -1001699979466       2      Marco     Aceti  acetimarco   \n",
       "68973   26170256 -1001157628331       2      Marco     Aceti  acetimarco   \n",
       "38242   26170256 -1001167949464       2      Marco     Aceti  acetimarco   \n",
       "\n",
       "                                                    title  \n",
       "12116                               Chat | Studenti UniMi  \n",
       "19169            Algoritmi e strutture dati - Informatica  \n",
       "49409               Matematica del continuo - Informatica  \n",
       "23258   Ingegneria del software [PSS + PMD] (Club del ...  \n",
       "161472     Lingue e letterature straniere | StudentiUniMi  \n",
       "157771             International students | StudentiUniMi  \n",
       "68973   Programmazione 1 - Informatica, Informatica Mu...  \n",
       "38242                             Alloggi - StudentiUniMi  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged_walks = walk_results[walk_results.start > 0]\\\n",
    "                        .rename({\"start\":\"user_id\", \"ends\":\"group_id\", \"index\":\"visits\"}, axis=1)\\\n",
    "                        .merge(users)\\\n",
    "                        .merge(groups)\n",
    "\n",
    "merged_walks[merged_walks.username==\"acetimarco\"].sort_values(by=\"visits\", ascending=False)[:8]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In order to recommend a group to a user, I take the most visited groups (selecting between the ones in which he isn't already)."
   ]
  },
  {
   "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>group_id</th>\n",
       "      <th>visits</th>\n",
       "      <th>first_name</th>\n",
       "      <th>last_name</th>\n",
       "      <th>username</th>\n",
       "      <th>title</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>23258</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001379726039</td>\n",
       "      <td>3</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Ingegneria del software [PSS + PMD] (Club del ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38242</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001167949464</td>\n",
       "      <td>2</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Alloggi - StudentiUniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4820</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001530261207</td>\n",
       "      <td>1</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Natural interaction - Informatica magistrale</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>106280</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001300632521</td>\n",
       "      <td>1</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Crittografia 1 - Informatica | Informatica mus...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>146924</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001534900689</td>\n",
       "      <td>1</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Biotecnologie mediche | StudentiUniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>136532</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001215728849</td>\n",
       "      <td>1</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Fondamenti di social media digitali (Informati...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124971</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001552980805</td>\n",
       "      <td>1</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Programmazione avanzata - Informatica magistrale</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115443</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001226553523</td>\n",
       "      <td>1</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Principi e modelli della percezione - Informat...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>111649</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001524770629</td>\n",
       "      <td>1</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Metodi probabilistici per l'informatica - Info...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>109232</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001212847374</td>\n",
       "      <td>1</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Matematica del Continuo - Informatica per la c...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         user_id       group_id  visits first_name last_name    username  \\\n",
       "23258   26170256 -1001379726039       3      Marco     Aceti  acetimarco   \n",
       "38242   26170256 -1001167949464       2      Marco     Aceti  acetimarco   \n",
       "4820    26170256 -1001530261207       1      Marco     Aceti  acetimarco   \n",
       "106280  26170256 -1001300632521       1      Marco     Aceti  acetimarco   \n",
       "146924  26170256 -1001534900689       1      Marco     Aceti  acetimarco   \n",
       "136532  26170256 -1001215728849       1      Marco     Aceti  acetimarco   \n",
       "124971  26170256 -1001552980805       1      Marco     Aceti  acetimarco   \n",
       "115443  26170256 -1001226553523       1      Marco     Aceti  acetimarco   \n",
       "111649  26170256 -1001524770629       1      Marco     Aceti  acetimarco   \n",
       "109232  26170256 -1001212847374       1      Marco     Aceti  acetimarco   \n",
       "\n",
       "                                                    title  \n",
       "23258   Ingegneria del software [PSS + PMD] (Club del ...  \n",
       "38242                             Alloggi - StudentiUniMi  \n",
       "4820         Natural interaction - Informatica magistrale  \n",
       "106280  Crittografia 1 - Informatica | Informatica mus...  \n",
       "146924              Biotecnologie mediche | StudentiUniMi  \n",
       "136532  Fondamenti di social media digitali (Informati...  \n",
       "124971   Programmazione avanzata - Informatica magistrale  \n",
       "115443  Principi e modelli della percezione - Informat...  \n",
       "111649  Metodi probabilistici per l'informatica - Info...  \n",
       "109232  Matematica del Continuo - Informatica per la c...  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def get_not_in_groups(user_id):\n",
    "    in_groups = group_membership[group_membership.user_id == user_id].group_id\n",
    "    not_in_groups = groups[~(\n",
    "        groups.group_id.isin(in_groups))].group_id\n",
    "    return not_in_groups\n",
    "\n",
    "def get_top_k(user, k):\n",
    "    return merged_walks[(merged_walks.user_id==user) & (merged_walks.group_id.isin(get_not_in_groups(user)))]\\\n",
    "        .sort_values(by=\"visits\", ascending=False)[:k]\n",
    "\n",
    "get_top_k(26170256, 10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To restrict the field, instead of working with the whole graph, I create a subgraph for each user (based on the groups degree id and the supposed degree id of the user) and perform the random walks from it. The subgraph contains the groups of the degree and the extra groups."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-1001563734995    [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...\n",
       "-1001557200491    [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...\n",
       "-1001774201871    [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.6, ...\n",
       "-1001724030284    [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...\n",
       "-1001531478970    [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...\n",
       "dtype: object"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "group_description = pd.read_pickle(\"group_description.pkl\")\n",
    "group_description.head()"
   ]
  },
  {
   "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>degree_id</th>\n",
       "      <th>degree_name</th>\n",
       "      <th>degree_type</th>\n",
       "      <th>degree_group_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>66</th>\n",
       "      <td>1</td>\n",
       "      <td>Informatica</td>\n",
       "      <td>B</td>\n",
       "      <td>-1001437343087</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    degree_id  degree_name degree_type  degree_group_id\n",
       "66          1  Informatica           B   -1001437343087"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "group_degrees = pd.read_pickle(\"groups_degrees.pkl\")\n",
    "trainset = pd.read_pickle(\"trainset.pkl\")\n",
    "\n",
    "#returns -1 if no probable degree is detected and 158 for FOR24 groups\n",
    "def get_probable_degree_id(user_id):\n",
    "    in_groups = trainset[trainset.user_id == user_id].group_id\n",
    "    features_sum = sum(group_description[in_groups])[:158]\n",
    "    if sum(features_sum) == 0:\n",
    "        return -1\n",
    "    return np.argmax(features_sum)+1\n",
    "\n",
    "degrees = pd.read_pickle(\"degrees.pkl\")\n",
    "degrees[degrees.degree_id==get_probable_degree_id(26170256)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_user_subgraph(user_id):\n",
    "    degree_id = get_probable_degree_id(user_id)\n",
    "    \n",
    "    if degree_id == -1:\n",
    "        subgroups = group_degrees[(group_degrees.degree_id.isna())].group_id.values\n",
    "    else: \n",
    "        subgroups = group_degrees[(group_degrees.degree_id.isna()) | (\n",
    "            group_degrees.degree_id == degree_id)].group_id.values\n",
    "\n",
    "    sub_group_membership = trainset[trainset.group_id.isin(\n",
    "        subgroups)]\n",
    "        \n",
    "    assert len(sub_group_membership[sub_group_membership.user_id == user_id]) != 0\n",
    "    \n",
    "    G = nx.Graph()\n",
    "    for _,row in sub_group_membership.iterrows():\n",
    "        G.add_edge(row.user_id, row.group_id, weight=row.messages_count)\n",
    "    return G\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_graph_predictions(user_id, walklen=6, epochs=50, return_weight=1, neighbor_weight=1):\n",
    "    \n",
    "    user_subgraph = get_user_subgraph(user_id)\n",
    "\n",
    "    G = cg.csrgraph(user_subgraph)\n",
    "    node_names = G.names\n",
    "    random_walks = G.random_walks(walklen=walklen, epochs=epochs, \\\n",
    "        return_weight=return_weight, neighbor_weight=neighbor_weight)\n",
    "\n",
    "    labeled_walks = np.vectorize(lambda x : node_names[x])(random_walks)\n",
    "\n",
    "    start_end_points = list()\n",
    "    for walk in labeled_walks:\n",
    "        start_end_points.append((walk[0], walk[len(walk)-1]))\n",
    "\n",
    "    walk_results = pd.DataFrame(start_end_points)\\\n",
    "                .rename({0:\"start\", 1:\"ends\"}, axis=1)\\\n",
    "                .reset_index()\\\n",
    "                .groupby([\"start\",\"ends\"])\\\n",
    "                .count()\\\n",
    "                .reset_index()\n",
    "\n",
    "    merged_walks = walk_results[walk_results.start > 0]\\\n",
    "                            .rename({\"start\":\"user_id\", \"ends\":\"group_id\", \"index\":\"prediction\"}, axis=1)\\\n",
    "                            .merge(users)\\\n",
    "                            .merge(groups)\n",
    "    return merged_walks\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>group_id</th>\n",
       "      <th>prediction</th>\n",
       "      <th>first_name</th>\n",
       "      <th>last_name</th>\n",
       "      <th>username</th>\n",
       "      <th>title</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001350552358</td>\n",
       "      <td>2</td>\n",
       "      <td>Alessia</td>\n",
       "      <td>NaN</td>\n",
       "      <td>aleceres</td>\n",
       "      <td>Intelligent Systems for Industry, Supply Chain...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001181636991</td>\n",
       "      <td>2</td>\n",
       "      <td>Alessia</td>\n",
       "      <td>NaN</td>\n",
       "      <td>aleceres</td>\n",
       "      <td>Test di Inglese - StudentiUniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001348261542</td>\n",
       "      <td>2</td>\n",
       "      <td>Alessia</td>\n",
       "      <td>NaN</td>\n",
       "      <td>aleceres</td>\n",
       "      <td>Modellazione e analisi di sistemi - Informatic...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001260565113</td>\n",
       "      <td>1</td>\n",
       "      <td>Alessia</td>\n",
       "      <td>NaN</td>\n",
       "      <td>aleceres</td>\n",
       "      <td>Compravendita libri e appunti - StudentiUniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001167949464</td>\n",
       "      <td>1</td>\n",
       "      <td>Alessia</td>\n",
       "      <td>NaN</td>\n",
       "      <td>aleceres</td>\n",
       "      <td>Alloggi - StudentiUniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001220140042</td>\n",
       "      <td>1</td>\n",
       "      <td>Alessia</td>\n",
       "      <td>NaN</td>\n",
       "      <td>aleceres</td>\n",
       "      <td>Advent Of Code | StudentiUniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001288175726</td>\n",
       "      <td>1</td>\n",
       "      <td>Alessia</td>\n",
       "      <td>NaN</td>\n",
       "      <td>aleceres</td>\n",
       "      <td>Visione artificiale - Informatica magistrale</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>350313104</td>\n",
       "      <td>-1001452073794</td>\n",
       "      <td>1</td>\n",
       "      <td>Alessia</td>\n",
       "      <td>NaN</td>\n",
       "      <td>aleceres</td>\n",
       "      <td>Statistical methods for machine learning - Inf...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     user_id       group_id  prediction first_name last_name  username  \\\n",
       "0  350313104 -1001350552358           2    Alessia       NaN  aleceres   \n",
       "1  350313104 -1001181636991           2    Alessia       NaN  aleceres   \n",
       "2  350313104 -1001348261542           2    Alessia       NaN  aleceres   \n",
       "3  350313104 -1001260565113           1    Alessia       NaN  aleceres   \n",
       "4  350313104 -1001167949464           1    Alessia       NaN  aleceres   \n",
       "5  350313104 -1001220140042           1    Alessia       NaN  aleceres   \n",
       "6  350313104 -1001288175726           1    Alessia       NaN  aleceres   \n",
       "7  350313104 -1001452073794           1    Alessia       NaN  aleceres   \n",
       "\n",
       "                                               title  \n",
       "0  Intelligent Systems for Industry, Supply Chain...  \n",
       "1                    Test di Inglese - StudentiUniMi  \n",
       "2  Modellazione e analisi di sistemi - Informatic...  \n",
       "3      Compravendita libri e appunti - StudentiUniMi  \n",
       "4                            Alloggi - StudentiUniMi  \n",
       "5                     Advent Of Code | StudentiUniMi  \n",
       "6       Visione artificiale - Informatica magistrale  \n",
       "7  Statistical methods for machine learning - Inf...  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def get_not_in_groups(user_id):\n",
    "    in_groups = trainset[trainset.user_id == user_id].group_id\n",
    "    not_in_groups = groups[~(\n",
    "        groups.group_id.isin(in_groups))].group_id\n",
    "    return not_in_groups\n",
    "\n",
    "def get_top_k_graph(user, k):\n",
    "    merged_walks = get_graph_predictions(user)\n",
    "    return merged_walks[(merged_walks.user_id==user) & (merged_walks.group_id.isin(get_not_in_groups(user)))]\\\n",
    "        .sort_values(by=\"prediction\", ascending=False)[:k]\n",
    "\n",
    "user_id = 350313104\n",
    "graph_predictions = get_top_k_graph(user_id, 8).reset_index().drop(columns=[\"index\"])\n",
    "graph_predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a9f5f6659bd04f07905fc518636cafcd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/4183 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# we parallelize in order to take less time in computing\n",
    "k = 8\n",
    "graphs = Parallel(n_jobs=-1, prefer=\"processes\")(\n",
    "    delayed(get_top_k_graph)(user, k) for user in tqdm(trainset.user_id.unique(), total=len(trainset.user_id.unique()))\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "graph_predictions = pd.DataFrame()\n",
    "\n",
    "for prediction in graphs:\n",
    "    graph_predictions = pd.concat([graph_predictions, prediction ])\n",
    "\n",
    "graph_predictions\n",
    "\n",
    "with open(\"predictions_graphs.pkl\", \"wb+\") as f:\n",
    "    pickle.dump(graph_predictions, f)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>group_id</th>\n",
       "      <th>prediction</th>\n",
       "      <th>first_name</th>\n",
       "      <th>last_name</th>\n",
       "      <th>username</th>\n",
       "      <th>title</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>16983</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001379726039</td>\n",
       "      <td>5</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Ingegneria del software [PSS + PMD] (Club del ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10588</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001440238500</td>\n",
       "      <td>3</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Reti di calcolatori - Informatica</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30304</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001181636991</td>\n",
       "      <td>3</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Test di Inglese - StudentiUniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7792</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001456212600</td>\n",
       "      <td>2</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Chat | Studenti UniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52563</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001213858785</td>\n",
       "      <td>2</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Ripetizioni - Studenti UniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2156</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001469541498</td>\n",
       "      <td>1</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Sicurezza e Privatezza - Informatica | Informa...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3679</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001466214340</td>\n",
       "      <td>1</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Elaborazione dei segnali - Informatica Musical...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35648</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001362716461</td>\n",
       "      <td>1</td>\n",
       "      <td>Marco</td>\n",
       "      <td>Aceti</td>\n",
       "      <td>acetimarco</td>\n",
       "      <td>Intelligenza Artificiale 1 - Informatica</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        user_id       group_id  prediction first_name last_name    username  \\\n",
       "16983  26170256 -1001379726039           5      Marco     Aceti  acetimarco   \n",
       "10588  26170256 -1001440238500           3      Marco     Aceti  acetimarco   \n",
       "30304  26170256 -1001181636991           3      Marco     Aceti  acetimarco   \n",
       "7792   26170256 -1001456212600           2      Marco     Aceti  acetimarco   \n",
       "52563  26170256 -1001213858785           2      Marco     Aceti  acetimarco   \n",
       "2156   26170256 -1001469541498           1      Marco     Aceti  acetimarco   \n",
       "3679   26170256 -1001466214340           1      Marco     Aceti  acetimarco   \n",
       "35648  26170256 -1001362716461           1      Marco     Aceti  acetimarco   \n",
       "\n",
       "                                                   title  \n",
       "16983  Ingegneria del software [PSS + PMD] (Club del ...  \n",
       "10588                  Reti di calcolatori - Informatica  \n",
       "30304                    Test di Inglese - StudentiUniMi  \n",
       "7792                               Chat | Studenti UniMi  \n",
       "52563                       Ripetizioni - Studenti UniMi  \n",
       "2156   Sicurezza e Privatezza - Informatica | Informa...  \n",
       "3679   Elaborazione dei segnali - Informatica Musical...  \n",
       "35648           Intelligenza Artificiale 1 - Informatica  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "graph_predictions[graph_predictions.user_id==26170256]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I compute precision and recall metrics on this implementation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "testset = pd.read_pickle(\"testset.pkl\")\n",
    "\n",
    "def is_relevant(user, item):\n",
    "    return len(\n",
    "            testset[\n",
    "                (testset.user_id == user) & (testset.group_id == item)\n",
    "            ]\n",
    "            ) != 0\n",
    "\n",
    "\n",
    "def HR(user, k):\n",
    "    recommended_items = graph_predictions[graph_predictions.user_id==user][:k]\n",
    "    return sum(is_relevant(user, item) for item in recommended_items.group_id)\n",
    "\n",
    "\n",
    "def average_precision(user, k):\n",
    "    recommended_items = graph_predictions[graph_predictions.user_id==user][:k]\n",
    "    return sum(\n",
    "        is_relevant(user, row[1].group_id) * (1 / (rank + 1) * 1)\n",
    "        for rank, row in enumerate(recommended_items.iterrows())\n",
    "    )\n",
    "\n",
    "def RR(user,k):\n",
    "    recommended_items = graph_predictions[graph_predictions.user_id==user][:k]\n",
    "    return (sum(is_relevant(user,row[1].group_id)*(1/(rank+1)) for rank,row in enumerate(recommended_items.iterrows())))\n",
    "\n",
    "def precision(user,k):\n",
    "    recommended_items = graph_predictions[graph_predictions.user_id==user][:k]\n",
    "    return sum(is_relevant(user,item) for item in recommended_items.group_id)*1/k"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average P@8: 0.059637278263171827\n",
      "HR@8: 0.4770982261053746\n",
      "MAP@8: 0.25094840954650327\n",
      "MRR@8: 0.25094840954650327\n"
     ]
    }
   ],
   "source": [
    "k=8\n",
    "print(f\"Average P@{k}: {np.mean([precision(user,k) for user in testset.user_id])}\")\n",
    "print(f\"HR@{k}: {np.mean([HR(user,k) for user in testset.user_id])}\")\n",
    "print(f\"MAP@{k}: {np.mean([average_precision(user,k) for user in testset.user_id])}\")\n",
    "print(f\"MRR@{k}: {np.mean([RR(user,k) for user in testset.user_id])}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then I try to combine the outputs of the content based recommender system (implemented in contentbased.ipynb) with the graph one (implemented here). To do it, I use the reciprocal rank method: based on the position n in the final ranking, each group has a rank of 1/n. For each group in the first and second recommendation, I sum the two ranks and take the ones with the highest value."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "content_based_predictions = pd.read_pickle(\"all_merged_predictions.pkl\")\n",
    "graph_predictions_rank = graph_predictions.reset_index()\n",
    "graph_predictions_rank.drop(columns=[\"index\"], inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_reciprocal_rank(user, group, predictions):\n",
    "    return 1/len(predictions) - \\\n",
    "                 predictions[(predictions.user_id == user) & \\\n",
    "                                   (predictions.group_id == group)]\\\n",
    "                 .index[0] + 1 \n",
    "\n",
    "user = content_based_predictions.user_id[0]\n",
    "reciprocal_rank_predictions = list()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>group_id</th>\n",
       "      <th>pred</th>\n",
       "      <th>title</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001379726039</td>\n",
       "      <td>-652.999970</td>\n",
       "      <td>Ingegneria del software [PSS + PMD] (Club del ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001440238500</td>\n",
       "      <td>-653.999970</td>\n",
       "      <td>Reti di calcolatori - Informatica</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001181636991</td>\n",
       "      <td>-654.999970</td>\n",
       "      <td>Test di Inglese - StudentiUniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001456212600</td>\n",
       "      <td>-655.999970</td>\n",
       "      <td>Chat | Studenti UniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001213858785</td>\n",
       "      <td>-656.999970</td>\n",
       "      <td>Ripetizioni - Studenti UniMi</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001362716461</td>\n",
       "      <td>-659.999970</td>\n",
       "      <td>Intelligenza Artificiale 1 - Informatica</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001189502801</td>\n",
       "      <td>-619243.000000</td>\n",
       "      <td>Linguaggi e traduttori - Informatica | Informa...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>26170256</td>\n",
       "      <td>-1001466214340</td>\n",
       "      <td>-908957.999969</td>\n",
       "      <td>Elaborazione dei segnali - Informatica Musical...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    user_id       group_id           pred  \\\n",
       "0  26170256 -1001379726039    -652.999970   \n",
       "1  26170256 -1001440238500    -653.999970   \n",
       "2  26170256 -1001181636991    -654.999970   \n",
       "3  26170256 -1001456212600    -655.999970   \n",
       "4  26170256 -1001213858785    -656.999970   \n",
       "5  26170256 -1001362716461    -659.999970   \n",
       "6  26170256 -1001189502801 -619243.000000   \n",
       "7  26170256 -1001466214340 -908957.999969   \n",
       "\n",
       "                                               title  \n",
       "0  Ingegneria del software [PSS + PMD] (Club del ...  \n",
       "1                  Reti di calcolatori - Informatica  \n",
       "2                    Test di Inglese - StudentiUniMi  \n",
       "3                              Chat | Studenti UniMi  \n",
       "4                       Ripetizioni - Studenti UniMi  \n",
       "5           Intelligenza Artificiale 1 - Informatica  \n",
       "6  Linguaggi e traduttori - Informatica | Informa...  \n",
       "7  Elaborazione dei segnali - Informatica Musical...  "
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user = 26170256\n",
    "k=8\n",
    "\n",
    "def get_rr_recommendation(user, k):\n",
    "    top_k_graph = graph_predictions_rank[graph_predictions_rank.user_id==user]\n",
    "    top_k_cb = content_based_predictions[content_based_predictions.user_id==user]\\\n",
    "        .sort_values(by= [\"prediction\"], ascending=False)[:k]\n",
    "\n",
    "    reciprocal_rank_predictions = list()\n",
    "    for group in set(list(top_k_graph.group_id.values) + list(top_k_cb.group_id.values)):\n",
    "        rank = 0 \n",
    "        if group in top_k_graph.group_id.values:\n",
    "            rank+= get_reciprocal_rank(user, group, graph_predictions_rank)\n",
    "        if group in top_k_cb.group_id.values:\n",
    "            rank+= get_reciprocal_rank(user, group, content_based_predictions)\n",
    "        reciprocal_rank_predictions.append((user, group, rank))\n",
    "\n",
    "    return pd.DataFrame(reciprocal_rank_predictions)\\\n",
    "            .rename({0:\"user_id\", 1:\"group_id\", 2:\"pred\"}, axis=1)\\\n",
    "            .sort_values(by= [\"pred\"], ascending=False)\\\n",
    "            .merge(groups)[:k]\n",
    "\n",
    "get_rr_recommendation(user, k)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "testset = pd.read_pickle(\"testset.pkl\")\n",
    "\n",
    "def is_relevant(user, item):\n",
    "    return len(\n",
    "            testset[\n",
    "                (testset.user_id == user) & (testset.group_id == item)\n",
    "            ]\n",
    "            ) != 0\n",
    "\n",
    "\n",
    "def HR(user, k):\n",
    "    recommended_items = get_rr_recommendation(user, k)\n",
    "    return sum(is_relevant(user, item) for item in recommended_items.group_id)\n",
    "\n",
    "\n",
    "def average_precision(user, k):\n",
    "    recommended_items = get_rr_recommendation(user, k)\n",
    "    return sum(\n",
    "        is_relevant(user, row[1].group_id) * (1 / (rank + 1) * 1)\n",
    "        for rank, row in enumerate(recommended_items.iterrows())\n",
    "    )\n",
    "\n",
    "def RR(user,k):\n",
    "    recommended_items = get_rr_recommendation(user, k)\n",
    "    return (sum(is_relevant(user,row[1].group_id)*(1/(rank+1)) for rank,row in enumerate(recommended_items.iterrows())))\n",
    "\n",
    "def precision(user,k):\n",
    "    recommended_items = get_rr_recommendation(user, k)\n",
    "    return sum(is_relevant(user,item) for item in recommended_items.group_id)*1/k"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average P@8: 0.059604183214191156\n",
      "HR@8: 0.47683346571352925\n"
     ]
    }
   ],
   "source": [
    "k=8\n",
    "print(f\"Average P@{k}: {np.mean([precision(user,k) for user in testset.user_id])}\")\n",
    "print(f\"HR@{k}: {np.mean([HR(user,k) for user in testset.user_id])}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAP@8: 0.3268797987821022\n",
      "MRR@8: 0.3268797987821022\n"
     ]
    }
   ],
   "source": [
    "print(f\"MAP@{k}: {np.mean([average_precision(user,k) for user in testset.user_id])}\")\n",
    "print(f\"MRR@{k}: {np.mean([RR(user,k) for user in testset.user_id])}\")"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "916dbcbb3f70747c44a77c7bcd40155683ae19c65e1c03b4aa3499c5328201f1"
  },
  "kernelspec": {
   "display_name": "Python 3.9.7 64-bit",
   "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": "3.9.7"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}