P6 sujet insatisfaction 0.1.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "81a8e4ba",
   "metadata": {},
   "source": [
    "# installation des librairies manquantes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "92399a40",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting spacy\n",
      "  Downloading spacy-3.2.1-cp38-cp38-win_amd64.whl (12.2 MB)\n",
      "Collecting wasabi<1.1.0,>=0.8.1\n",
      "  Downloading wasabi-0.9.0-py3-none-any.whl (25 kB)\n",
      "Collecting thinc<8.1.0,>=8.0.12\n",
      "  Downloading thinc-8.0.13-cp38-cp38-win_amd64.whl (1.0 MB)\n",
      "Collecting cymem<2.1.0,>=2.0.2\n",
      "  Downloading cymem-2.0.6-cp38-cp38-win_amd64.whl (36 kB)\n",
      "Requirement already satisfied: jinja2 in c:\\anaconda3\\lib\\site-packages (from spacy) (2.11.3)\n",
      "Requirement already satisfied: setuptools in c:\\anaconda3\\lib\\site-packages (from spacy) (52.0.0.post20210125)\n",
      "Collecting preshed<3.1.0,>=3.0.2\n",
      "  Downloading preshed-3.0.6-cp38-cp38-win_amd64.whl (113 kB)\n",
      "Collecting spacy-loggers<2.0.0,>=1.0.0\n",
      "  Downloading spacy_loggers-1.0.1-py3-none-any.whl (7.0 kB)\n",
      "Collecting spacy-legacy<3.1.0,>=3.0.8\n",
      "  Downloading spacy_legacy-3.0.8-py2.py3-none-any.whl (14 kB)\n",
      "Requirement already satisfied: numpy>=1.15.0 in c:\\anaconda3\\lib\\site-packages (from spacy) (1.21.5)\n",
      "Requirement already satisfied: tqdm<5.0.0,>=4.38.0 in c:\\anaconda3\\lib\\site-packages (from spacy) (4.59.0)\n",
      "Requirement already satisfied: requests<3.0.0,>=2.13.0 in c:\\anaconda3\\lib\\site-packages (from spacy) (2.25.1)\n",
      "Requirement already satisfied: packaging>=20.0 in c:\\anaconda3\\lib\\site-packages (from spacy) (20.9)\n",
      "Collecting catalogue<2.1.0,>=2.0.6\n",
      "  Downloading catalogue-2.0.6-py3-none-any.whl (17 kB)\n",
      "Collecting pathy>=0.3.5\n",
      "  Downloading pathy-0.6.1-py3-none-any.whl (42 kB)\n",
      "Collecting blis<0.8.0,>=0.4.0\n",
      "  Downloading blis-0.7.5-cp38-cp38-win_amd64.whl (6.6 MB)\n",
      "Collecting typer<0.5.0,>=0.3.0\n",
      "  Downloading typer-0.4.0-py3-none-any.whl (27 kB)\n",
      "Collecting langcodes<4.0.0,>=3.2.0\n",
      "  Downloading langcodes-3.3.0-py3-none-any.whl (181 kB)\n",
      "Collecting srsly<3.0.0,>=2.4.1\n",
      "  Downloading srsly-2.4.2-cp38-cp38-win_amd64.whl (452 kB)\n",
      "Collecting pydantic!=1.8,!=1.8.1,<1.9.0,>=1.7.4\n",
      "  Downloading pydantic-1.8.2-cp38-cp38-win_amd64.whl (2.0 MB)\n",
      "Collecting murmurhash<1.1.0,>=0.28.0\n",
      "  Downloading murmurhash-1.0.6-cp38-cp38-win_amd64.whl (21 kB)\n",
      "Requirement already satisfied: pyparsing>=2.0.2 in c:\\anaconda3\\lib\\site-packages (from packaging>=20.0->spacy) (2.4.7)\n",
      "Requirement already satisfied: smart-open<6.0.0,>=5.0.0 in c:\\anaconda3\\lib\\site-packages (from pathy>=0.3.5->spacy) (5.2.1)\n",
      "Requirement already satisfied: typing-extensions>=3.7.4.3 in c:\\users\\wesh\\appdata\\roaming\\python\\python38\\site-packages (from pydantic!=1.8,!=1.8.1,<1.9.0,>=1.7.4->spacy) (3.10.0.2)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in c:\\anaconda3\\lib\\site-packages (from requests<3.0.0,>=2.13.0->spacy) (2020.12.5)\n",
      "Requirement already satisfied: idna<3,>=2.5 in c:\\anaconda3\\lib\\site-packages (from requests<3.0.0,>=2.13.0->spacy) (2.10)\n",
      "Requirement already satisfied: urllib3<1.27,>=1.21.1 in c:\\anaconda3\\lib\\site-packages (from requests<3.0.0,>=2.13.0->spacy) (1.26.4)\n",
      "Requirement already satisfied: chardet<5,>=3.0.2 in c:\\anaconda3\\lib\\site-packages (from requests<3.0.0,>=2.13.0->spacy) (3.0.4)\n",
      "Requirement already satisfied: click<9.0.0,>=7.1.1 in c:\\anaconda3\\lib\\site-packages (from typer<0.5.0,>=0.3.0->spacy) (7.1.2)\n",
      "Requirement already satisfied: MarkupSafe>=0.23 in c:\\anaconda3\\lib\\site-packages (from jinja2->spacy) (1.1.1)\n",
      "Installing collected packages: murmurhash, cymem, catalogue, wasabi, typer, srsly, pydantic, preshed, blis, thinc, spacy-loggers, spacy-legacy, pathy, langcodes, spacy\n",
      "Successfully installed blis-0.7.5 catalogue-2.0.6 cymem-2.0.6 langcodes-3.3.0 murmurhash-1.0.6 pathy-0.6.1 preshed-3.0.6 pydantic-1.8.2 spacy-3.2.1 spacy-legacy-3.0.8 spacy-loggers-1.0.1 srsly-2.4.2 thinc-8.0.13 typer-0.4.0 wasabi-0.9.0\n"
     ]
    }
   ],
   "source": [
    "#bibliotheque langue\n",
    "import sys\n",
    "#!{sys.executable} -m pip install googletrans==3.1.0a0\n",
    "#!{sys.executable} -m pip install spacy\n",
    "#https://fasttext.cc/docs/en/supervised-tutorial.html\n",
    "# prerequis pour fastext: https://visualstudio.microsoft.com/fr/visual-cpp-build-tools/\n",
    "# tuto install visual c++ https://www.youtube.com/watch?v=IsAoIqnNia4\n",
    "#!{sys.executable} -m pip install fasttext\n",
    "# pour stockage feather binaries des dataframe\n",
    "#!{sys.executable} -m pip install pyarrow"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ff3fb9b2",
   "metadata": {},
   "source": [
    "# dezippage"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "01bafcc3",
   "metadata": {},
   "outputs": [],
   "source": [
    "#dezipage\n",
    "import shutil\n",
    "shutil.unpack_archive('input/raw/yelp_dataset.tgz','input/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "45a1a3d2",
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "import json\n",
    "\n",
    "#fonction interessante mais erreur quand on tombe sur un caractère non reconnu\n",
    "#au premier readline\n",
    "#donc à ne pas utiliser mais on garde pour la comprehension et quelques tricks\n",
    "def random_filter_json_toneveruse(filename,cut='proportion',p=0.01):    \n",
    "    filtered_data = []\n",
    "    length = -1\n",
    "    #with statement : simplifie l'accés à un fichier \n",
    "    # 'r'->read\n",
    "    with open(filename, 'r', encoding='utf-8') as f:\n",
    "        #enumerate est un iterateur\n",
    "        #permet de lire ligne à ligne de gros fichiers python\n",
    "        #le trick de la commande pass permet de recup le nombre de lignes\n",
    "        #la derniere valeur de length=nb lignes -1\n",
    "\n",
    "        #nombre de lignes\n",
    "        for length, l in enumerate(f):\n",
    "            pass\n",
    "        \n",
    "        #nombre de caractères\n",
    "        f.seek(0,2) #on pointe à la fin du fichier\n",
    "        nbchar=f.tell() #on retrouve la valeur du pointeur de caractère\n",
    "        \n",
    "        print(p)\n",
    "        print(length)\n",
    "        print(nbchar)\n",
    "        #nombre de lignes souhaitées\n",
    "        countline = list(range(int(p*length)+1))\n",
    "\n",
    "\n",
    "        for i in countline:\n",
    "            #tirage aléatoire du pointeur de caractère\n",
    "            ptchar=random.randint(0, nbchar)\n",
    "            \n",
    "            print(i,ptchar)\n",
    "            #on pointe sur le caractère tiré\n",
    "            #on sera tres probablement au milieu d'une ligne\n",
    "            f.seek(ptchar,0)\n",
    "                         \n",
    "            #on lis une premiere fois la ligne pour rien\n",
    "            #elle n'est pas entière\n",
    "            #mais ca nous deplace le pointeur de caractère à la prochaine ligne\n",
    "            \n",
    "            print(i,f.tell())\n",
    "            f.readline()\n",
    "            print(i,f.tell())\n",
    "            \n",
    "            #maintenant que le pointeur de caractère est idéalement positionné\n",
    "            #on peut lire la ligne\n",
    "            data_dict = json.loads(f.readline())\n",
    "            filtered_data.append(data_dict)\n",
    "            \n",
    "    \n",
    "    return loc_filtered_data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a26ab52f",
   "metadata": {},
   "source": [
    "# fonction de telechargement de gros json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "fc0e27fa",
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "import random\n",
    "import json\n",
    "import pandas as pd\n",
    "\n",
    "def random_filter_json(filename, p=0.01, typ='random', nbl=0):\n",
    "    '''\n",
    "    function which load partially or not a json file by line\n",
    "    in different manner\n",
    "        randomly or not:\n",
    "        typ :'random','first','last'\n",
    "        \n",
    "        p between 0 and 1 : keeped proportion if p=1 -> full loading\n",
    "            or\n",
    "        nbl: keeped lines\n",
    "        \n",
    "    '''\n",
    "    \n",
    "    start = time.time()\n",
    "    filtered_data = []\n",
    "    with open(filename, 'r', encoding='utf-8') as f:\n",
    "        \n",
    "        #calul du nombre de lignes dans le fichier\n",
    "        # ca peut prendre quelques secondes à cause de la boucle\n",
    "        length=sum(1 for line in f)\n",
    "\n",
    "        #calcul de la proportion de lignes à garder\n",
    "        if nbl>0:\n",
    "            p=min(1,nbl/length)\n",
    "            \n",
    "        \n",
    "        if (p==1):\n",
    "            f.seek(0,0)\n",
    "            for line in f:\n",
    "                data_dict = json.loads(line)\n",
    "                filtered_data.append(data_dict)\n",
    "    \n",
    "        elif (typ=='random' and p<1):\n",
    "            #on met le pointeur au debut du fchier par precaution\n",
    "            #si on a deja bouclé sur le fichier , le pointeur sera à la fin\n",
    "            #c est pour ca qu'on remet à 0\n",
    "            f.seek(0,0)\n",
    "            for line in f:\n",
    "                if random.random()<=p:\n",
    "                    data_dict = json.loads(line)\n",
    "                    filtered_data.append(data_dict)\n",
    "                \n",
    "        elif (typ=='first' and p<1):\n",
    "            j=-1\n",
    "            #on met le pointeur au debut du fchier par precaution\n",
    "            f.seek(0,0)\n",
    "            for line in f:\n",
    "                data_dict = json.loads(line)\n",
    "                filtered_data.append(data_dict)\n",
    "                j=j+1\n",
    "                if j>=(p*length):\n",
    "                    break\n",
    "                    \n",
    "        elif (typ=='last' and p<1):\n",
    "            j=-1\n",
    "            #on met le pointeur au debut du fchier par precaution\n",
    "            f.seek(0,0)\n",
    "            for line in f:\n",
    "                j=j+1\n",
    "                if j>=int((length-(p*length))):\n",
    "                    data_dict = json.loads(line)\n",
    "                    filtered_data.append(data_dict)\n",
    "\n",
    "\n",
    "            \n",
    "    filtered_dataframe=pd.DataFrame.from_dict(filtered_data)\n",
    "    print(time.time()-start)\n",
    "    return filtered_dataframe\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "299abb8d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "286.44788932800293\n",
      "60.36216497421265\n"
     ]
    }
   ],
   "source": [
    "#293 sec pour loading\n",
    "review=random_filter_json('input/yelp_academic_dataset_review.json',\n",
    "                   p=1, typ='first')\n",
    "#56 sec\n",
    "user=random_filter_json('input/yelp_academic_dataset_user.json',\n",
    "                        p=1, typ='first')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4007f9f4",
   "metadata": {},
   "source": [
    "# fonction de telechargement de petit json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "db4bbc5a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "def import_json(input,showhead=True, chunks=0):\n",
    "    '''import json file to pandas dataframe\n",
    "    \n",
    "        if chunks=0, all the json file is directly readed in a dataframe\n",
    "        if chunks>0, the json file is splited into several dataframes of chunks lines\n",
    "        they will be merged at the end\n",
    "        \n",
    "        larger is chunks, faster is the loading but memory can be saturated\n",
    "        smaller is chunks, better is the the chance to load all the file \n",
    "        \n",
    "        take into account that dataframe object is heavy and even this function\n",
    "        can do the job, it s not the most efficient for the large file in terms\n",
    "        of time\n",
    "    '''\n",
    "    start = time.time()\n",
    "    if chunks==0:\n",
    "        output = pd.read_json(input, lines=True)\n",
    "    else:\n",
    "        iterator=pd.read_json(input, lines=True, chunksize=chunks)\n",
    "        #initialisation du dataframe\n",
    "        output=pd.dataframe()\n",
    "        for chunk in iterator:\n",
    "            output=output.append(chunk)\n",
    "            \n",
    "    print(time.time()-start)\n",
    "    \n",
    "    if showhead==True:\n",
    "        display(output.head(5))\n",
    "    \n",
    "    return output\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "f9c0e08a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.5575804710388184\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>business_id</th>\n",
       "      <th>name</th>\n",
       "      <th>address</th>\n",
       "      <th>city</th>\n",
       "      <th>state</th>\n",
       "      <th>postal_code</th>\n",
       "      <th>latitude</th>\n",
       "      <th>longitude</th>\n",
       "      <th>stars</th>\n",
       "      <th>review_count</th>\n",
       "      <th>is_open</th>\n",
       "      <th>attributes</th>\n",
       "      <th>categories</th>\n",
       "      <th>hours</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6iYb2HFDywm3zjuRg0shjw</td>\n",
       "      <td>Oskar Blues Taproom</td>\n",
       "      <td>921 Pearl St</td>\n",
       "      <td>Boulder</td>\n",
       "      <td>CO</td>\n",
       "      <td>80302</td>\n",
       "      <td>40.017544</td>\n",
       "      <td>-105.283348</td>\n",
       "      <td>4.0</td>\n",
       "      <td>86</td>\n",
       "      <td>1</td>\n",
       "      <td>{'RestaurantsTableService': 'True', 'WiFi': 'u...</td>\n",
       "      <td>Gastropubs, Food, Beer Gardens, Restaurants, B...</td>\n",
       "      <td>{'Monday': '11:0-23:0', 'Tuesday': '11:0-23:0'...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>tCbdrRPZA0oiIYSmHG3J0w</td>\n",
       "      <td>Flying Elephants at PDX</td>\n",
       "      <td>7000 NE Airport Way</td>\n",
       "      <td>Portland</td>\n",
       "      <td>OR</td>\n",
       "      <td>97218</td>\n",
       "      <td>45.588906</td>\n",
       "      <td>-122.593331</td>\n",
       "      <td>4.0</td>\n",
       "      <td>126</td>\n",
       "      <td>1</td>\n",
       "      <td>{'RestaurantsTakeOut': 'True', 'RestaurantsAtt...</td>\n",
       "      <td>Salad, Soup, Sandwiches, Delis, Restaurants, C...</td>\n",
       "      <td>{'Monday': '5:0-18:0', 'Tuesday': '5:0-17:0', ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>bvN78flM8NLprQ1a1y5dRg</td>\n",
       "      <td>The Reclaimory</td>\n",
       "      <td>4720 Hawthorne Ave</td>\n",
       "      <td>Portland</td>\n",
       "      <td>OR</td>\n",
       "      <td>97214</td>\n",
       "      <td>45.511907</td>\n",
       "      <td>-122.613693</td>\n",
       "      <td>4.5</td>\n",
       "      <td>13</td>\n",
       "      <td>1</td>\n",
       "      <td>{'BusinessAcceptsCreditCards': 'True', 'Restau...</td>\n",
       "      <td>Antiques, Fashion, Used, Vintage &amp; Consignment...</td>\n",
       "      <td>{'Thursday': '11:0-18:0', 'Friday': '11:0-18:0...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>oaepsyvc0J17qwi8cfrOWg</td>\n",
       "      <td>Great Clips</td>\n",
       "      <td>2566 Enterprise Rd</td>\n",
       "      <td>Orange City</td>\n",
       "      <td>FL</td>\n",
       "      <td>32763</td>\n",
       "      <td>28.914482</td>\n",
       "      <td>-81.295979</td>\n",
       "      <td>3.0</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>{'RestaurantsPriceRange2': '1', 'BusinessAccep...</td>\n",
       "      <td>Beauty &amp; Spas, Hair Salons</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>PE9uqAjdw0E4-8mjGl3wVA</td>\n",
       "      <td>Crossfit Terminus</td>\n",
       "      <td>1046 Memorial Dr SE</td>\n",
       "      <td>Atlanta</td>\n",
       "      <td>GA</td>\n",
       "      <td>30316</td>\n",
       "      <td>33.747027</td>\n",
       "      <td>-84.353424</td>\n",
       "      <td>4.0</td>\n",
       "      <td>14</td>\n",
       "      <td>1</td>\n",
       "      <td>{'GoodForKids': 'False', 'BusinessParking': '{...</td>\n",
       "      <td>Gyms, Active Life, Interval Training Gyms, Fit...</td>\n",
       "      <td>{'Monday': '16:0-19:0', 'Tuesday': '16:0-19:0'...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              business_id                     name              address  \\\n",
       "0  6iYb2HFDywm3zjuRg0shjw      Oskar Blues Taproom         921 Pearl St   \n",
       "1  tCbdrRPZA0oiIYSmHG3J0w  Flying Elephants at PDX  7000 NE Airport Way   \n",
       "2  bvN78flM8NLprQ1a1y5dRg           The Reclaimory   4720 Hawthorne Ave   \n",
       "3  oaepsyvc0J17qwi8cfrOWg              Great Clips   2566 Enterprise Rd   \n",
       "4  PE9uqAjdw0E4-8mjGl3wVA        Crossfit Terminus  1046 Memorial Dr SE   \n",
       "\n",
       "          city state postal_code   latitude   longitude  stars  review_count  \\\n",
       "0      Boulder    CO       80302  40.017544 -105.283348    4.0            86   \n",
       "1     Portland    OR       97218  45.588906 -122.593331    4.0           126   \n",
       "2     Portland    OR       97214  45.511907 -122.613693    4.5            13   \n",
       "3  Orange City    FL       32763  28.914482  -81.295979    3.0             8   \n",
       "4      Atlanta    GA       30316  33.747027  -84.353424    4.0            14   \n",
       "\n",
       "   is_open                                         attributes  \\\n",
       "0        1  {'RestaurantsTableService': 'True', 'WiFi': 'u...   \n",
       "1        1  {'RestaurantsTakeOut': 'True', 'RestaurantsAtt...   \n",
       "2        1  {'BusinessAcceptsCreditCards': 'True', 'Restau...   \n",
       "3        1  {'RestaurantsPriceRange2': '1', 'BusinessAccep...   \n",
       "4        1  {'GoodForKids': 'False', 'BusinessParking': '{...   \n",
       "\n",
       "                                          categories  \\\n",
       "0  Gastropubs, Food, Beer Gardens, Restaurants, B...   \n",
       "1  Salad, Soup, Sandwiches, Delis, Restaurants, C...   \n",
       "2  Antiques, Fashion, Used, Vintage & Consignment...   \n",
       "3                         Beauty & Spas, Hair Salons   \n",
       "4  Gyms, Active Life, Interval Training Gyms, Fit...   \n",
       "\n",
       "                                               hours  \n",
       "0  {'Monday': '11:0-23:0', 'Tuesday': '11:0-23:0'...  \n",
       "1  {'Monday': '5:0-18:0', 'Tuesday': '5:0-17:0', ...  \n",
       "2  {'Thursday': '11:0-18:0', 'Friday': '11:0-18:0...  \n",
       "3                                               None  \n",
       "4  {'Monday': '16:0-19:0', 'Tuesday': '16:0-19:0'...  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.2492001056671143\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>business_id</th>\n",
       "      <th>date</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>--0r8K_AQ4FZfLsX3ZYRDA</td>\n",
       "      <td>2017-09-03 17:13:59</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>--0zrn43LEaB4jUWTQH_Bg</td>\n",
       "      <td>2010-10-08 22:21:20, 2010-11-01 21:29:14, 2010...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>--164t1nclzzmca7eDiJMw</td>\n",
       "      <td>2010-02-26 02:06:53, 2010-02-27 08:00:09, 2010...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>--2aF9NhXnNVpDV0KS3xBQ</td>\n",
       "      <td>2014-11-03 16:35:35, 2015-01-30 18:16:03, 2015...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>--2mEJ63SC_8_08_jGgVIg</td>\n",
       "      <td>2010-12-15 17:10:46, 2013-12-28 00:27:54, 2015...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              business_id                                               date\n",
       "0  --0r8K_AQ4FZfLsX3ZYRDA                                2017-09-03 17:13:59\n",
       "1  --0zrn43LEaB4jUWTQH_Bg  2010-10-08 22:21:20, 2010-11-01 21:29:14, 2010...\n",
       "2  --164t1nclzzmca7eDiJMw  2010-02-26 02:06:53, 2010-02-27 08:00:09, 2010...\n",
       "3  --2aF9NhXnNVpDV0KS3xBQ  2014-11-03 16:35:35, 2015-01-30 18:16:03, 2015...\n",
       "4  --2mEJ63SC_8_08_jGgVIg  2010-12-15 17:10:46, 2013-12-28 00:27:54, 2015..."
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6.702863454818726\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user_id</th>\n",
       "      <th>business_id</th>\n",
       "      <th>text</th>\n",
       "      <th>date</th>\n",
       "      <th>compliment_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>WCjg0jdHXMlwbqS9tZUx8Q</td>\n",
       "      <td>ENwBByjpoa5Gg7tKgxqwLg</td>\n",
       "      <td>Carne asada chips...</td>\n",
       "      <td>2011-07-22 19:07:35</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>42-Z02y9bABShAGZhuSzrQ</td>\n",
       "      <td>jKO4Og6ucdX2-YCTKQVYjg</td>\n",
       "      <td>Best happy hour from 3pm to 6pm! $1 off martin...</td>\n",
       "      <td>2014-09-10 07:33:29</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5u7E3LYp_3eB8dLuUBazXQ</td>\n",
       "      <td>9Bto7mky640ocgezVKSfVg</td>\n",
       "      <td>Nice people, skilled staff, clean location - b...</td>\n",
       "      <td>2013-12-13 23:23:41</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>wDWoMG5N9oI4DJ-p7z8EBg</td>\n",
       "      <td>XWFjKtRGZ9khRGtGg2ZvaA</td>\n",
       "      <td>1/2-price bowling &amp; the \"Very\" Old Fashion are...</td>\n",
       "      <td>2017-07-11 23:07:16</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>JmuFlorjjRshHTKzTwNtgg</td>\n",
       "      <td>mkrx0VhSMU3p3uhyJGCoWA</td>\n",
       "      <td>Solid gold's. Great sauna. Great staff, too. E...</td>\n",
       "      <td>2016-11-30 08:46:36</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  user_id             business_id  \\\n",
       "0  WCjg0jdHXMlwbqS9tZUx8Q  ENwBByjpoa5Gg7tKgxqwLg   \n",
       "1  42-Z02y9bABShAGZhuSzrQ  jKO4Og6ucdX2-YCTKQVYjg   \n",
       "2  5u7E3LYp_3eB8dLuUBazXQ  9Bto7mky640ocgezVKSfVg   \n",
       "3  wDWoMG5N9oI4DJ-p7z8EBg  XWFjKtRGZ9khRGtGg2ZvaA   \n",
       "4  JmuFlorjjRshHTKzTwNtgg  mkrx0VhSMU3p3uhyJGCoWA   \n",
       "\n",
       "                                                text                date  \\\n",
       "0                               Carne asada chips... 2011-07-22 19:07:35   \n",
       "1  Best happy hour from 3pm to 6pm! $1 off martin... 2014-09-10 07:33:29   \n",
       "2  Nice people, skilled staff, clean location - b... 2013-12-13 23:23:41   \n",
       "3  1/2-price bowling & the \"Very\" Old Fashion are... 2017-07-11 23:07:16   \n",
       "4  Solid gold's. Great sauna. Great staff, too. E... 2016-11-30 08:46:36   \n",
       "\n",
       "   compliment_count  \n",
       "0                 0  \n",
       "1                 0  \n",
       "2                 0  \n",
       "3                 0  \n",
       "4                 0  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import time\n",
    "import random\n",
    "import json\n",
    "import pandas as pd\n",
    "\n",
    "business=import_json('input/yelp_academic_dataset_business.json')\n",
    "checkin=import_json('input/yelp_academic_dataset_checkin.json')\n",
    "tip=import_json('input/yelp_academic_dataset_tip.json')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "990a1524",
   "metadata": {},
   "source": [
    "# petite analyse des business\n",
    "* distribution des notes des reviews\n",
    "* distribution du nombre de reviews"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "1f2abd5c-f697-4785-b193-71ce49a6dc0b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5       0.086210\n",
       "6       0.069894\n",
       "7       0.058667\n",
       "8       0.048878\n",
       "9       0.041305\n",
       "          ...   \n",
       "991     0.000006\n",
       "1199    0.000006\n",
       "1455    0.000006\n",
       "1583    0.000006\n",
       "1919    0.000006\n",
       "Name: review_count, Length: 1281, dtype: float64"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('default') #specially for dark theme\n",
    "\n",
    "pie1=business.stars.value_counts(dropna=False)\n",
    "pie1=pd.DataFrame(pie1)\n",
    "\n",
    "business.review_count.value_counts(normalize=True)\n",
    "\n",
    "plt.rcParams[\"figure.figsize\"] = [8, 6]\n",
    "plt.rcParams[\"figure.autolayout\"] = True\n",
    "f, axes = plt.subplots(1, 1)\n",
    "\n",
    "axes.pie(pie1.stars, \n",
    "         labels=pie1.index, \n",
    "         normalize=True,\n",
    "         autopct='%1.1f%%',\n",
    "         shadow=True,\n",
    "         startangle=90)\n",
    "\n",
    "display(business.review_count.value_counts(normalize=True))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "98b6f93c",
   "metadata": {},
   "source": [
    "# detection de langage\n",
    "* googletrans trop lent\n",
    "* fastext rapide comme l eclair"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2b077fae",
   "metadata": {},
   "outputs": [],
   "source": [
    "#https://stackoverflow.com/questions/39142778/python-how-to-determine-the-language\n",
    "#https://stackoverflow.com/questions/52455774/googletrans-stopped-working-with-error-nonetype-object-has-no-attribute-group\n",
    "#https://medium.com/analytics-vidhya/translate-list-and-pandas-data-frame-using-googletrans-library-in-python-f28b8cb84f21\n",
    "from googletrans import Translator\n",
    "import time\n",
    "#import googletrans\n",
    "#print(googletrans.__version__)\n",
    "\n",
    "def goog_rev_lang_detect(datain=review.iloc[0:10].copy()):\n",
    "    trans=Translator()\n",
    "\n",
    "    rev=datain\n",
    "    #rev['text2']=rev['text'].str[0:100]\n",
    "\n",
    "    #detection de la langue via google trans\n",
    "    #5 en 1.7 sec (0.32 dans l aprem)\n",
    "    #50 en 20 sec (2.63 dans l aprem)\n",
    "    #150 en 60 sec (60 dans l aprem, j'en deduis qu il y a un ralentisseur apres x requetes)\n",
    "    # donc pour 22700 lignes on va tourner en 150 min , c est long\n",
    "    #raccorucir le texte ne fait pas gagner de temps\n",
    "\n",
    "    #polyglot et fasttext(facebook) doivent etre plus rapide et sont donc à tester\n",
    "    #https://polyglot.readthedocs.io/en/latest/Installation.html\n",
    "    #polyglot un peu chiant à installer à cause des dependances instables en windows\n",
    "\n",
    "    start = time.time()\n",
    "    rev['trans']=rev['text'].apply(trans.detect)\n",
    "    print(time.time()-start)\n",
    "\n",
    "    #separation des attributs renvoyés\n",
    "    rev['lang']=rev['trans'].apply(getattr,args=('lang',))\n",
    "    rev['conf']=rev['trans'].apply(getattr,args=('confidence',))\n",
    "\n",
    "    #display(rev['lang'].value_counts(normalize=True))\n",
    "    return rev\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8b0c1f19",
   "metadata": {},
   "outputs": [],
   "source": [
    "#detection de langage fast\n",
    "#https://fasttext.cc/docs/en/language-identification.html\n",
    "#https://amitness.com/2019/07/identify-text-language-python/\n",
    "import fasttext\n",
    "\n",
    "#chargement complet 254 sec\n",
    "review=random_filter_json('input/yelp_academic_dataset_review.json',\n",
    "                   p=1, typ='random',nbl=0)\n",
    "\n",
    "rev=review.copy()\n",
    "\n",
    "#detection de la langue via fasttext\n",
    "#150 en 0.02 sec: c'est rapide\n",
    "#6.28 sec pour 100000 lignes c est trés rapide\n",
    "#600 sec /10 min pour les plus de 8 millions de lignes\n",
    "\n",
    "PRETRAINED_MODEL_PATH = 'fasttext_lib_lgid/lid.176.bin'\n",
    "model = fasttext.load_model(PRETRAINED_MODEL_PATH)\n",
    "#on enlève les sauts de ligne\n",
    "rev['text2']=rev['text'].str.replace(\"\\n\",\"\")\n",
    "\n",
    "start = time.time()\n",
    "#petit trick pour unpack le tuple fourni par modèl predict\n",
    "rev[['lang','conf']]=pd.DataFrame(rev['text2'].apply(model.predict).tolist())\n",
    "#on prend la premiere valeur des tuple\n",
    "rev['lang']=rev['lang'].str[0]\n",
    "rev['conf']=rev['conf'].str[0]\n",
    "#on nettoie la langue\n",
    "rev['lang']=rev['lang'].str.replace(r\"^__label__\",\"\")\n",
    "print(time.time()-start)\n",
    "\n",
    "display(rev['lang'].value_counts(normalize=True))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1dd08175",
   "metadata": {},
   "source": [
    "# sauvegarde en binaries des gros dataframes\n",
    "* pour etre plus efficace et sauvegarder\n",
    "* feather semble la plus rapide : dependance pyarrow"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "aca1b656",
   "metadata": {},
   "outputs": [],
   "source": [
    "#https://towardsdatascience.com/the-best-format-to-save-pandas-data-414dca023e0d\n",
    "#feather à privilegier\n",
    "\n",
    "import time\n",
    "import pandas as pd\n",
    "\n",
    "#sauvegarde\n",
    "start = time.time()\n",
    "rev.to_feather('input/rev1')\n",
    "print(time.time()-start)\n",
    "\n",
    "#DL\n",
    "\n",
    "start = time.time()\n",
    "rev1=pd.read_feather('input/rev1')\n",
    "print(time.time()-start)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5a5e7397",
   "metadata": {},
   "source": [
    "# filtrage des reviews"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "53e856dd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "temps DL review bin: 0.0\n",
      "2.8537752628326416\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>business_id</th>\n",
       "      <th>name</th>\n",
       "      <th>address</th>\n",
       "      <th>city</th>\n",
       "      <th>state</th>\n",
       "      <th>postal_code</th>\n",
       "      <th>latitude</th>\n",
       "      <th>longitude</th>\n",
       "      <th>stars</th>\n",
       "      <th>review_count</th>\n",
       "      <th>is_open</th>\n",
       "      <th>attributes</th>\n",
       "      <th>categories</th>\n",
       "      <th>hours</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6iYb2HFDywm3zjuRg0shjw</td>\n",
       "      <td>Oskar Blues Taproom</td>\n",
       "      <td>921 Pearl St</td>\n",
       "      <td>Boulder</td>\n",
       "      <td>CO</td>\n",
       "      <td>80302</td>\n",
       "      <td>40.017544</td>\n",
       "      <td>-105.283348</td>\n",
       "      <td>4.0</td>\n",
       "      <td>86</td>\n",
       "      <td>1</td>\n",
       "      <td>{'RestaurantsTableService': 'True', 'WiFi': 'u...</td>\n",
       "      <td>Gastropubs, Food, Beer Gardens, Restaurants, B...</td>\n",
       "      <td>{'Monday': '11:0-23:0', 'Tuesday': '11:0-23:0'...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>tCbdrRPZA0oiIYSmHG3J0w</td>\n",
       "      <td>Flying Elephants at PDX</td>\n",
       "      <td>7000 NE Airport Way</td>\n",
       "      <td>Portland</td>\n",
       "      <td>OR</td>\n",
       "      <td>97218</td>\n",
       "      <td>45.588906</td>\n",
       "      <td>-122.593331</td>\n",
       "      <td>4.0</td>\n",
       "      <td>126</td>\n",
       "      <td>1</td>\n",
       "      <td>{'RestaurantsTakeOut': 'True', 'RestaurantsAtt...</td>\n",
       "      <td>Salad, Soup, Sandwiches, Delis, Restaurants, C...</td>\n",
       "      <td>{'Monday': '5:0-18:0', 'Tuesday': '5:0-17:0', ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>bvN78flM8NLprQ1a1y5dRg</td>\n",
       "      <td>The Reclaimory</td>\n",
       "      <td>4720 Hawthorne Ave</td>\n",
       "      <td>Portland</td>\n",
       "      <td>OR</td>\n",
       "      <td>97214</td>\n",
       "      <td>45.511907</td>\n",
       "      <td>-122.613693</td>\n",
       "      <td>4.5</td>\n",
       "      <td>13</td>\n",
       "      <td>1</td>\n",
       "      <td>{'BusinessAcceptsCreditCards': 'True', 'Restau...</td>\n",
       "      <td>Antiques, Fashion, Used, Vintage &amp; Consignment...</td>\n",
       "      <td>{'Thursday': '11:0-18:0', 'Friday': '11:0-18:0...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>oaepsyvc0J17qwi8cfrOWg</td>\n",
       "      <td>Great Clips</td>\n",
       "      <td>2566 Enterprise Rd</td>\n",
       "      <td>Orange City</td>\n",
       "      <td>FL</td>\n",
       "      <td>32763</td>\n",
       "      <td>28.914482</td>\n",
       "      <td>-81.295979</td>\n",
       "      <td>3.0</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>{'RestaurantsPriceRange2': '1', 'BusinessAccep...</td>\n",
       "      <td>Beauty &amp; Spas, Hair Salons</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>PE9uqAjdw0E4-8mjGl3wVA</td>\n",
       "      <td>Crossfit Terminus</td>\n",
       "      <td>1046 Memorial Dr SE</td>\n",
       "      <td>Atlanta</td>\n",
       "      <td>GA</td>\n",
       "      <td>30316</td>\n",
       "      <td>33.747027</td>\n",
       "      <td>-84.353424</td>\n",
       "      <td>4.0</td>\n",
       "      <td>14</td>\n",
       "      <td>1</td>\n",
       "      <td>{'GoodForKids': 'False', 'BusinessParking': '{...</td>\n",
       "      <td>Gyms, Active Life, Interval Training Gyms, Fit...</td>\n",
       "      <td>{'Monday': '16:0-19:0', 'Tuesday': '16:0-19:0'...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              business_id                     name              address  \\\n",
       "0  6iYb2HFDywm3zjuRg0shjw      Oskar Blues Taproom         921 Pearl St   \n",
       "1  tCbdrRPZA0oiIYSmHG3J0w  Flying Elephants at PDX  7000 NE Airport Way   \n",
       "2  bvN78flM8NLprQ1a1y5dRg           The Reclaimory   4720 Hawthorne Ave   \n",
       "3  oaepsyvc0J17qwi8cfrOWg              Great Clips   2566 Enterprise Rd   \n",
       "4  PE9uqAjdw0E4-8mjGl3wVA        Crossfit Terminus  1046 Memorial Dr SE   \n",
       "\n",
       "          city state postal_code   latitude   longitude  stars  review_count  \\\n",
       "0      Boulder    CO       80302  40.017544 -105.283348    4.0            86   \n",
       "1     Portland    OR       97218  45.588906 -122.593331    4.0           126   \n",
       "2     Portland    OR       97214  45.511907 -122.613693    4.5            13   \n",
       "3  Orange City    FL       32763  28.914482  -81.295979    3.0             8   \n",
       "4      Atlanta    GA       30316  33.747027  -84.353424    4.0            14   \n",
       "\n",
       "   is_open                                         attributes  \\\n",
       "0        1  {'RestaurantsTableService': 'True', 'WiFi': 'u...   \n",
       "1        1  {'RestaurantsTakeOut': 'True', 'RestaurantsAtt...   \n",
       "2        1  {'BusinessAcceptsCreditCards': 'True', 'Restau...   \n",
       "3        1  {'RestaurantsPriceRange2': '1', 'BusinessAccep...   \n",
       "4        1  {'GoodForKids': 'False', 'BusinessParking': '{...   \n",
       "\n",
       "                                          categories  \\\n",
       "0  Gastropubs, Food, Beer Gardens, Restaurants, B...   \n",
       "1  Salad, Soup, Sandwiches, Delis, Restaurants, C...   \n",
       "2  Antiques, Fashion, Used, Vintage & Consignment...   \n",
       "3                         Beauty & Spas, Hair Salons   \n",
       "4  Gyms, Active Life, Interval Training Gyms, Fit...   \n",
       "\n",
       "                                               hours  \n",
       "0  {'Monday': '11:0-23:0', 'Tuesday': '11:0-23:0'...  \n",
       "1  {'Monday': '5:0-18:0', 'Tuesday': '5:0-17:0', ...  \n",
       "2  {'Thursday': '11:0-18:0', 'Friday': '11:0-18:0...  \n",
       "3                                               None  \n",
       "4  {'Monday': '16:0-19:0', 'Tuesday': '16:0-19:0'...  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "temps jointure review et review_count de business: 227.84871697425842\n"
     ]
    }
   ],
   "source": [
    "import time\n",
    "import pandas as pd\n",
    "import json\n",
    "\n",
    "#DL reviews\n",
    "start = time.time()\n",
    "rev1=pd.read_feather('input/rev1')\n",
    "print('temps DL review bin:',time.time()-start)\n",
    "\n",
    "#DL business\n",
    "business=import_json('input/yelp_academic_dataset_business.json')\n",
    "\n",
    "#mise en place du meme index et jointure\n",
    "start = time.time()\n",
    "business=business.set_index('business_id')\n",
    "rev1=rev1.set_index('business_id')\n",
    "rev1=rev1.join(business.review_count)\n",
    "print('temps jointure review et review_count de business:',time.time()-start)\n",
    "\n",
    "#dict unpack\n",
    "#business = pd.concat([business, business['attributes'].apply(pd.Series)], axis=1)\n",
    "#print(business.review_count.mean())\n",
    "#print(business.review_count.median())\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "5552c227-2953-4b11-b53a-4e4940065dea",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4CxF8c3MB7VAdY8zFb2cZQ    9295\n",
       "OQ2oHkcWA8KNC1Lsvj1SBA    7394\n",
       "PrsvO1rzkgg6qFizlAoEtg    7385\n",
       "zioLxtBc9THNS2TOn9xW1w    6325\n",
       "y2w6rFaO0XEiG5mFfOsiFA    5220\n",
       "                          ... \n",
       "m1o8TRiZp7ACWiLl6MyEmg     699\n",
       "7yiTxAYqUKt2UVs4ceK57g     699\n",
       "mxyCf3UlVPLFeQEDOZrphw     699\n",
       "HIhgfQtM9KXpjj1iZATxiQ     699\n",
       "gUHpQYwW_fd0l0hcE2i6Dg     698\n",
       "Name: business_id, Length: 1000, dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rev1.index.value_counts().head(1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "e303c685",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "temps de filtrage mauvaise note & nb de reviews: 77.3948585987091\n",
      "9.002398252487183\n"
     ]
    }
   ],
   "source": [
    "#filtrage\n",
    "    #reviews > 698\n",
    "    #language =en\n",
    "    \n",
    "start = time.time()\n",
    "#revfilt=rev1.loc[(rev1.stars<3)]\n",
    "revfilt=rev1.loc[(rev1.review_count>698)] #les 1000 plus gros restaurants\n",
    "revfilt=revfilt.loc[(revfilt.lang=='en') & (revfilt.conf>0.95)]\n",
    "print('temps de filtrage:',time.time()-start)\n",
    "\n",
    "#sauvegarde\n",
    "start = time.time()\n",
    "revfilt=revfilt.reset_index()\n",
    "revfilt.to_feather('input/revfilt')\n",
    "print(time.time()-start)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c0ffe667-9d91-4150-820a-f0c427e220ab",
   "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": "3.8.8"
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}