POI_category_prediction/ML_Models.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "3ad89bc2",
   "metadata": {},
   "source": [
    "## Predicting the POI category from the name of it "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "d83c5f61",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'D:\\\\MSC\\\\Data'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os \n",
    "os.chdir('D:\\MSC\\Data')\n",
    "os.getcwd()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "ca4c0cd2",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np \n",
    "import pandas as pd \n",
    "import re\n",
    "import matplotlib.pyplot as plt\n",
    "from nltk.tokenize import regexp_tokenize\n",
    "import re"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "2352e915",
   "metadata": {},
   "outputs": [],
   "source": [
    "places = pd.read_csv('places\\\\places for dss\\\\thi_jpura_dehi_cleaned_places.csv',index_col = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "5d069e3b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(29261, 12)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "places.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "3a8e41f2",
   "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>name</th>\n",
       "      <th>place_id</th>\n",
       "      <th>lat</th>\n",
       "      <th>lng</th>\n",
       "      <th>types</th>\n",
       "      <th>vicinity</th>\n",
       "      <th>no_of_ratings</th>\n",
       "      <th>avg_rating</th>\n",
       "      <th>geometry</th>\n",
       "      <th>type_1</th>\n",
       "      <th>dual_1</th>\n",
       "      <th>dual_2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Khairiya Muslim Girl's Grand College</td>\n",
       "      <td>ChIJS6PgLw9Z4joRADXcRN0MYh0</td>\n",
       "      <td>6.929464</td>\n",
       "      <td>79.868975</td>\n",
       "      <td>['school', 'point_of_interest', 'establishment']</td>\n",
       "      <td>219 Sri Vajiragnana Mawatha, Colombo</td>\n",
       "      <td>27.0</td>\n",
       "      <td>4.4</td>\n",
       "      <td>POINT (79.8689747 6.929464200000002)</td>\n",
       "      <td>school</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>My abacus centre</td>\n",
       "      <td>ChIJsVF2JQ9Z4joRIe5Tj4_xfJs</td>\n",
       "      <td>6.929437</td>\n",
       "      <td>79.868935</td>\n",
       "      <td>['school', 'point_of_interest', 'establishment']</td>\n",
       "      <td>Sri Vajiragnana Mawatha, Colombo</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>POINT (79.8689347 6.929437099999999)</td>\n",
       "      <td>school</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Sea and Air Shipping Colombo</td>\n",
       "      <td>ChIJf02bZ5pZ4joRFjvKh9SKx78</td>\n",
       "      <td>6.929457</td>\n",
       "      <td>79.868527</td>\n",
       "      <td>['point_of_interest', 'establishment']</td>\n",
       "      <td>D 1/3,, Sri Vagiraghana Mawatha I, Colombo 10</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>POINT (79.868527 6.929457)</td>\n",
       "      <td>establishment</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Asian Pharmacy</td>\n",
       "      <td>ChIJTf5fsQhZ4joRRc4Z_cyWnFA</td>\n",
       "      <td>6.929506</td>\n",
       "      <td>79.869235</td>\n",
       "      <td>['pharmacy', 'health', 'point_of_interest', 's...</td>\n",
       "      <td>228 Sri Vajiragnana Mawatha, Colombo</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>POINT (79.86923519999999 6.929506300000001)</td>\n",
       "      <td>pharmacy</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>LAUGFS SuperMart</td>\n",
       "      <td>ChIJ4TBnybtZ4joRICKdUnmmLfw</td>\n",
       "      <td>6.929335</td>\n",
       "      <td>79.869333</td>\n",
       "      <td>['supermarket', 'grocery_or_supermarket', 'foo...</td>\n",
       "      <td>No.228 Sri Vajiragnana Mawatha, Colombo</td>\n",
       "      <td>2.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>POINT (79.8693329 6.929334599999998)</td>\n",
       "      <td>supermarket</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                   name                     place_id  \\\n",
       "0  Khairiya Muslim Girl's Grand College  ChIJS6PgLw9Z4joRADXcRN0MYh0   \n",
       "1                      My abacus centre  ChIJsVF2JQ9Z4joRIe5Tj4_xfJs   \n",
       "2          Sea and Air Shipping Colombo  ChIJf02bZ5pZ4joRFjvKh9SKx78   \n",
       "4                        Asian Pharmacy  ChIJTf5fsQhZ4joRRc4Z_cyWnFA   \n",
       "5                      LAUGFS SuperMart  ChIJ4TBnybtZ4joRICKdUnmmLfw   \n",
       "\n",
       "        lat        lng                                              types  \\\n",
       "0  6.929464  79.868975   ['school', 'point_of_interest', 'establishment']   \n",
       "1  6.929437  79.868935   ['school', 'point_of_interest', 'establishment']   \n",
       "2  6.929457  79.868527             ['point_of_interest', 'establishment']   \n",
       "4  6.929506  79.869235  ['pharmacy', 'health', 'point_of_interest', 's...   \n",
       "5  6.929335  79.869333  ['supermarket', 'grocery_or_supermarket', 'foo...   \n",
       "\n",
       "                                        vicinity  no_of_ratings  avg_rating  \\\n",
       "0           219 Sri Vajiragnana Mawatha, Colombo           27.0         4.4   \n",
       "1               Sri Vajiragnana Mawatha, Colombo            NaN         NaN   \n",
       "2  D 1/3,, Sri Vagiraghana Mawatha I, Colombo 10            NaN         NaN   \n",
       "4           228 Sri Vajiragnana Mawatha, Colombo            NaN         NaN   \n",
       "5        No.228 Sri Vajiragnana Mawatha, Colombo            2.0         5.0   \n",
       "\n",
       "                                      geometry         type_1 dual_1 dual_2  \n",
       "0         POINT (79.8689747 6.929464200000002)         school    NaN    NaN  \n",
       "1         POINT (79.8689347 6.929437099999999)         school    NaN    NaN  \n",
       "2                   POINT (79.868527 6.929457)  establishment    NaN    NaN  \n",
       "4  POINT (79.86923519999999 6.929506300000001)       pharmacy    NaN    NaN  \n",
       "5         POINT (79.8693329 6.929334599999998)    supermarket    NaN    NaN  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "places.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "5b336be6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['name', 'place_id', 'lat', 'lng', 'types', 'vicinity', 'no_of_ratings',\n",
       "       'avg_rating', 'geometry', 'type_1', 'dual_1', 'dual_2'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "places.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d51f7c4c",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-warning\">\n",
    "<b> Text Cleaning :</b> \n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "b7f9ab13",
   "metadata": {},
   "outputs": [],
   "source": [
    "def isEnglish(s):\n",
    "    try:\n",
    "        s.encode(encoding='utf-8').decode('ascii')\n",
    "    except UnicodeDecodeError:\n",
    "        return False\n",
    "    else:\n",
    "        return True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "8b470eff",
   "metadata": {},
   "outputs": [],
   "source": [
    "## tokenize and lower the poi names\n",
    "def name_clean(name):\n",
    "    # Define a regex pattern to tokenize: pattern1\n",
    "    pattern1 = r\"\\w+\"\n",
    "    # Use the pattern to tokenize\n",
    "    words = regexp_tokenize(name, pattern1)\n",
    "#     words = re.findall('[A-Za-z]' , words)\n",
    "    l_words = [t.lower() for t in words if t.isalpha() and isEnglish(t) and len(t)>1]\n",
    "    length = len(l_words)\n",
    "    if length == 0: \n",
    "        l_words = ['NA']\n",
    "    return ' '.join(l_words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "6b57b5b3",
   "metadata": {},
   "outputs": [],
   "source": [
    "places['clean_name'] = places['name'].apply(name_clean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "fd89986c",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# unreliable_names_indexes = []\n",
    "# for idx,row in places.iterrows():\n",
    "#     letters =  re.findall('[A-Za-z]' ,  row['name'])\n",
    "#     if len(letters)<3:\n",
    "#         unreliable_names_indexes.append(idx)\n",
    "# named_places = places.drop(unreliable_names_indexes)\n",
    "# named_places.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "9cec6f37",
   "metadata": {},
   "outputs": [],
   "source": [
    "named_places = places[places['clean_name']!='NA'] ## remove the POIS without accceptable names "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "b7e566cd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(28949, 13)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "named_places.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "f5aa67b3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(28928, 13)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## remove the dual type of POIs\n",
    "named_places = named_places.drop(named_places[named_places['type_1']=='dual'].index)\n",
    "named_places.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e7106a14",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-warning\">\n",
    "<b> Add features for word count : do not use now for this analysis </b> \n",
    "</div>"
   ]
  },
  {
   "cell_type": "raw",
   "id": "3c843c8a",
   "metadata": {},
   "source": [
    "named_places['letter_length'] = named_places['clean_name'].apply(len)"
   ]
  },
  {
   "cell_type": "raw",
   "id": "fd1a3659",
   "metadata": {},
   "source": [
    "def word_count(word):\n",
    "    words = word.split()\n",
    "    return len(words)"
   ]
  },
  {
   "cell_type": "raw",
   "id": "4a4829a8",
   "metadata": {},
   "source": [
    "named_places['words_length'] = named_places['clean_name'].apply(word_count)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aace402d",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-warning\">\n",
    "<b> Category assignment  :</b> \n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "2d901d21",
   "metadata": {},
   "outputs": [],
   "source": [
    "categories = pd.read_csv('places\\categories\\categories_summary_of_places.csv',index_col=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "4a8b695d",
   "metadata": {},
   "outputs": [],
   "source": [
    "## small function to remove the nan values in the above cateogry arrays\n",
    "def arraywithoutnan(nanarray):\n",
    "    newarray = []\n",
    "    for i in range(len(nanarray)):\n",
    "        if str(nanarray[i]) != 'nan':\n",
    "            newarray.append(nanarray[i])\n",
    "        if str(nanarray[i]) == 'nan':\n",
    "            break\n",
    "    return newarray "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "f731bcb2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# create arrays for each category type removing the nan values \n",
    "shopping = arraywithoutnan(categories.loc['shopping ',:].values)\n",
    "education = arraywithoutnan(categories.loc['Education ',:].values)\n",
    "medical = arraywithoutnan(categories.loc['medical',:].values)\n",
    "recreational = arraywithoutnan(categories.loc['recreational ',:].values)\n",
    "personal = arraywithoutnan(categories.loc['personal ',:].values)\n",
    "transit = arraywithoutnan(categories.loc['transit ',:].values)\n",
    "dining = arraywithoutnan(categories.loc['dining ',:].values)\n",
    "home = arraywithoutnan(categories.loc['home ',:].values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "f4f7ce07",
   "metadata": {},
   "outputs": [],
   "source": [
    "categories_list = [shopping,education,medical,recreational,personal,transit,dining,home]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "a25e8a4e",
   "metadata": {},
   "outputs": [],
   "source": [
    "## function to provide the category given the category type of a POI, \n",
    "def catassign(category_type):\n",
    "    if category_type == 'work':\n",
    "        categry = 'work'\n",
    "    elif category_type in shopping:\n",
    "        categry = 'shopping'\n",
    "    elif category_type in education: \n",
    "        categry = 'education'\n",
    "    elif category_type in medical: \n",
    "        categry = 'medical'\n",
    "    elif category_type in recreational:\n",
    "        categry = 'recreational'\n",
    "    elif category_type in personal:\n",
    "        categry = 'personal'\n",
    "    elif category_type in transit:\n",
    "        categry = 'transit'\n",
    "    elif category_type in dining: \n",
    "        categry = 'dining'\n",
    "    elif category_type in home: \n",
    "        categry = 'home'\n",
    "    else:\n",
    "        categry = 'unknown'\n",
    "    return categry "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "1030f3da",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "for idx,row in named_places.iterrows():\n",
    "     named_places.loc[idx,'purpose'] = catassign(row['type_1'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "76252bf9",
   "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>name</th>\n",
       "      <th>place_id</th>\n",
       "      <th>lat</th>\n",
       "      <th>lng</th>\n",
       "      <th>types</th>\n",
       "      <th>vicinity</th>\n",
       "      <th>no_of_ratings</th>\n",
       "      <th>avg_rating</th>\n",
       "      <th>geometry</th>\n",
       "      <th>type_1</th>\n",
       "      <th>dual_1</th>\n",
       "      <th>dual_2</th>\n",
       "      <th>clean_name</th>\n",
       "      <th>purpose</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [name, place_id, lat, lng, types, vicinity, no_of_ratings, avg_rating, geometry, type_1, dual_1, dual_2, clean_name, purpose]\n",
       "Index: []"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "named_places[(named_places['purpose'] == 'unknown') & (named_places['type_1'] != 'establishment')] ## this means alll the unkonwn catgories are relate to the establishment category type"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8a3b4d24",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-warning\">\n",
    "<b> ML Model Creation  :</b> \n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "88b98c7b",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "# from sklearn.model_selection import cross_val_score\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "# from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.metrics import classification_report \n",
    "from sklearn.metrics import accuracy_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "452f51ff",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.naive_bayes import MultinomialNB"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "71a05d5e",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer,TfidfTransformer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "9f1cd7c5",
   "metadata": {},
   "outputs": [],
   "source": [
    "## dimension reduction techniques \n",
    "from sklearn.decomposition import NMF\n",
    "from sklearn.decomposition import TruncatedSVD\n",
    "from sklearn.discriminant_analysis import LinearDiscriminantAnalysis\n",
    "from sklearn.decomposition import PCA, KernelPCA\n",
    "from sklearn.feature_selection import VarianceThreshold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "2697f1eb",
   "metadata": {},
   "outputs": [],
   "source": [
    "places_data = named_places[named_places['purpose'] != 'unknown'] # extract the labeled dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "7e92900c",
   "metadata": {},
   "outputs": [],
   "source": [
    "## select the categories to be predicted "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "a1fc2587",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['education', 'medical', 'shopping', 'personal', 'dining',\n",
       "       'recreational', 'transit', 'home'], dtype=object)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "places_data['purpose'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "7e90fcfa",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(15553, 14)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "places_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "4842f018",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "personal        5717\n",
       "shopping        4344\n",
       "dining          2396\n",
       "medical         1289\n",
       "education       1195\n",
       "transit          349\n",
       "recreational     252\n",
       "home              11\n",
       "Name: purpose, dtype: int64"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "places_data['purpose'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "8a38ee6e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAEuCAYAAABiY8IIAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAAAf0klEQVR4nO3de7hdVX3u8e9LUKRgBNpAaRIEPUEFBIUIUTzViqfGooa2YqNVUovNU4pWq7UNbU9Vaiy1jz0tekCpF4I3jNekWFSaeqmCYCJoCEhJgUJKhHglVQ8afM8fcyxYe2ftG+6sOfce7+d59rPWHHvNtX57J/tdc405xpiyTURE1GGvtguIiIjhSehHRFQkoR8RUZGEfkRERRL6EREVSehHRFRkUqEv6QBJH5H0DUk3SnqypIMkXSHp5nJ7YN/jz5G0VdJNkp7V136CpM3le+dL0p74oSIiYrDJHun/A/Ap248FjgNuBFYBG2wvAjaUbSQdBSwHjgaWAhdImlOe50JgJbCofC2dpp8jIiImYcLQlzQX+GXgXQC2f2z7e8AyYE152BrgtHJ/GXCp7Xtt3wpsBU6UdCgw1/ZVbmaEXdK3T0REDMFkjvQfBewA3iPpWknvlLQfcIjt7QDl9uDy+PnAHX37bytt88v90e0RETEke0/yMccDr7B9taR/oHTljGFQP73Had/9CaSVNN1A7Lfffic89rGPnUSZERHRs2nTpm/Znje6fTKhvw3YZvvqsv0RmtC/S9KhtreXrpu7+x6/sG//BcCdpX3BgPbd2L4IuAhg8eLF3rhx4yTKjIiIHkn/Oah9wu4d298E7pD0mNJ0CnADsB5YUdpWAOvK/fXAckn7SDqC5oTtNaULaKekJWXUzhl9+0RExBBM5kgf4BXA+yU9FLgFeCnNG8ZaSWcCtwOnA9jeImktzRvDLuBs2/eV5zkLuBjYF7i8fEVExJCo60srp3snImLqJG2yvXh0e2bkRkRUJKEfEVGRhH5EREUS+hERFUnoR0RUZLJDNjvv8FWfnPbnvO28U6f9OSMi2pQj/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIikwq9CXdJmmzpOskbSxtB0m6QtLN5fbAvsefI2mrpJskPauv/YTyPFslnS9J0/8jRUTEWKZypP8rtp9ge3HZXgVssL0I2FC2kXQUsBw4GlgKXCBpTtnnQmAlsKh8Lf3Zf4SIiJisn6V7ZxmwptxfA5zW136p7Xtt3wpsBU6UdCgw1/ZVtg1c0rdPREQMwWRD38BnJG2StLK0HWJ7O0C5Pbi0zwfu6Nt3W2mbX+6Pbo+IiCHZe5KPO9n2nZIOBq6Q9I1xHjuon97jtO/+BM0by0qAww47bJIlRkTERCZ1pG/7znJ7N/Bx4ETgrtJlQ7m9uzx8G7Cwb/cFwJ2lfcGA9kGvd5HtxbYXz5s3b/I/TUREjGvC0Je0n6SH9+4DvwpcD6wHVpSHrQDWlfvrgeWS9pF0BM0J22tKF9BOSUvKqJ0z+vaJiIghmEz3ziHAx8voyr2BD9j+lKSvAGslnQncDpwOYHuLpLXADcAu4Gzb95XnOgu4GNgXuLx8RUTEkEwY+rZvAY4b0P5t4JQx9lkNrB7QvhE4ZuplRkTEdMiM3IiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKjIZC6MHtPo8FWfnPbnvO28U6f9OSNidsqRfkRERRL6EREVSehHRFQkoR8RUZGEfkRERRL6EREVSehHRFQkoR8RUZFJh76kOZKulXRZ2T5I0hWSbi63B/Y99hxJWyXdJOlZfe0nSNpcvne+JE3vjxMREeOZypH+K4Eb+7ZXARtsLwI2lG0kHQUsB44GlgIXSJpT9rkQWAksKl9Lf6bqIyJiSiYV+pIWAKcC7+xrXgasKffXAKf1tV9q+17btwJbgRMlHQrMtX2VbQOX9O0TERFDMNkj/b8H/gT4aV/bIba3A5Tbg0v7fOCOvsdtK23zy/3R7buRtFLSRkkbd+zYMckSIyJiIhOGvqTnAHfb3jTJ5xzUT+9x2ndvtC+yvdj24nnz5k3yZSMiYiKTWWXzZOB5kn4NeBgwV9L7gLskHWp7e+m6ubs8fhuwsG//BcCdpX3BgPaIiBiSCY/0bZ9je4Htw2lO0P6r7RcD64EV5WErgHXl/npguaR9JB1Bc8L2mtIFtFPSkjJq54y+fSIiYgh+lvX0zwPWSjoTuB04HcD2FklrgRuAXcDZtu8r+5wFXAzsC1xeviIiYkimFPq2Pwd8rtz/NnDKGI9bDawe0L4ROGaqRUZExPTIjNyIiIok9CMiKpLQj4ioSEI/IqIiCf2IiIok9CMiKpLQj4ioSEI/IqIiCf2IiIok9CMiKpLQj4ioSEI/IqIiCf2IiIok9CMiKpLQj4ioSEI/IqIiCf2IiIok9CMiKpLQj4ioSEI/IqIiCf2IiIok9CMiKpLQj4ioSEI/IqIiCf2IiIok9CMiKpLQj4ioyIShL+lhkq6R9DVJWyS9obQfJOkKSTeX2wP79jlH0lZJN0l6Vl/7CZI2l++dL0l75seKiIhBJnOkfy/wDNvHAU8AlkpaAqwCNtheBGwo20g6ClgOHA0sBS6QNKc814XASmBR+Vo6fT9KRERMZMLQd+O/y+ZDypeBZcCa0r4GOK3cXwZcavte27cCW4ETJR0KzLV9lW0Dl/TtExERQzCpPn1JcyRdB9wNXGH7auAQ29sByu3B5eHzgTv6dt9W2uaX+6PbB73eSkkbJW3csWPHFH6ciIgYz6RC3/Z9tp8ALKA5aj9mnIcP6qf3OO2DXu8i24ttL543b95kSoyIiEmY0ugd298DPkfTF39X6bKh3N5dHrYNWNi32wLgztK+YEB7REQMyWRG78yTdEC5vy/wTOAbwHpgRXnYCmBdub8eWC5pH0lH0JywvaZ0Ae2UtKSM2jmjb5+IiBiCvSfxmEOBNWUEzl7AWtuXSboKWCvpTOB24HQA21skrQVuAHYBZ9u+rzzXWcDFwL7A5eUrIiKGZMLQt/114IkD2r8NnDLGPquB1QPaNwLjnQ+IiIg9KDNyIyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIikzmyllRocNXfXLan/O2806d9ueMiKnJkX5EREUS+hERFUnoR0RUJKEfEVGRhH5EREUS+hERFUnoR0RUJKEfEVGRhH5EREUS+hERFZkw9CUtlPRZSTdK2iLplaX9IElXSLq53B7Yt885krZKuknSs/raT5C0uXzvfEnaMz9WREQMMpkj/V3Aa2w/DlgCnC3pKGAVsMH2ImBD2aZ8bzlwNLAUuEDSnPJcFwIrgUXla+k0/iwRETGBCUPf9nbbXy33dwI3AvOBZcCa8rA1wGnl/jLgUtv32r4V2AqcKOlQYK7tq2wbuKRvn4iIGIIp9elLOhx4InA1cIjt7dC8MQAHl4fNB+7o221baZtf7o9uj4iIIZl06EvaH/go8Crb94z30AFtHqd90GutlLRR0sYdO3ZMtsSIiJjApEJf0kNoAv/9tj9Wmu8qXTaU27tL+zZgYd/uC4A7S/uCAe27sX2R7cW2F8+bN2+yP0tERExgMqN3BLwLuNH23/V9az2wotxfAazra18uaR9JR9CcsL2mdAHtlLSkPOcZfftERMQQTObKWScDLwE2S7qutP0ZcB6wVtKZwO3A6QC2t0haC9xAM/LnbNv3lf3OAi4G9gUuL18RETEkE4a+7S8yuD8e4JQx9lkNrB7QvhE4ZioFRkTE9MmM3IiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiIJ/YiIiiT0IyIqktCPiKhIQj8ioiITXhg9ossOX/XJaX/O2847ddqfM6IrcqQfEVGRhH5EREUS+hERFUmffsQQTPe5h5x3iAcrR/oRERVJ6EdEVGTC0Jf0bkl3S7q+r+0gSVdIurncHtj3vXMkbZV0k6Rn9bWfIGlz+d75kjT9P05ERIxnMkf6FwNLR7WtAjbYXgRsKNtIOgpYDhxd9rlA0pyyz4XASmBR+Rr9nBERsYdNGPq2vwB8Z1TzMmBNub8GOK2v/VLb99q+FdgKnCjpUGCu7atsG7ikb5+IiBiSB9unf4jt7QDl9uDSPh+4o+9x20rb/HJ/dHtERAzRdJ/IHdRP73HaBz+JtFLSRkkbd+zYMW3FRUTU7sGG/l2ly4Zye3dp3wYs7HvcAuDO0r5gQPtAti+yvdj24nnz5j3IEiMiYrQHG/rrgRXl/gpgXV/7ckn7SDqC5oTtNaULaKekJWXUzhl9+0RExJBMOCNX0geBpwO/IGkb8DrgPGCtpDOB24HTAWxvkbQWuAHYBZxt+77yVGfRjATaF7i8fEVExBBNGPq2XzjGt04Z4/GrgdUD2jcCx0ypuoiImFaZkRsRUZGEfkRERRL6EREVSehHRFQkoR8RUZGEfkRERRL6EREVSehHRFQkoR8RUZGEfkRERRL6EREVSehHRFQkoR8RUZGEfkRERRL6EREVSehHRFQkoR8RUZGEfkRERSa8XGJE1OHwVZ+c9ue87bxTp/0542eTI/2IiIok9CMiKpLQj4ioSPr0I2JGybmHn02O9CMiKpLQj4ioSEI/IqIiCf2IiIok9CMiKjL00Je0VNJNkrZKWjXs14+IqNlQQ1/SHOD/As8GjgJeKOmoYdYQEVGzYR/pnwhstX2L7R8DlwLLhlxDRES1ZHt4LyY9H1hq+2Vl+yXASbZfPupxK4GVZfMxwE3TXMovAN+a5uecbjOhRkid0y11Tq+a63yk7XmjG4c9I1cD2nZ717F9EXDRHitC2mh78Z56/ukwE2qE1DndUuf0Sp27G3b3zjZgYd/2AuDOIdcQEVGtYYf+V4BFko6Q9FBgObB+yDVERFRrqN07tndJejnwaWAO8G7bW4ZZQ7HHuo6m0UyoEVLndEud0yt1jjLUE7kREdGuzMiNiKhIQj8ioiIJ/YiIiiT0I2JCkk6eTFt036w+kSvp+PG+b/urw6plMiRtZvfJat8HNgJvtP3t4Ve1O0m/MaD5+8Bm23cPu57RJL2VAZP+emz/4RDLmZCkI4HXAo+kb0Sd7We0VtQokr5q+/iJ2to0xt8PNJNCbfvYIZc0JkkCfht4lO1zJR0G/KLta/b0a8/2a+S+ZZzvGejMH1VxOXAf8IGyvbzc3gNcDDy3hZoGORN4MvDZsv104MvAkZLOtf3etgorNrb8+lP1YeDtwD/S/Pt3hqQnA08B5kl6dd+35tIMu+6S57RdwBRcAPyUJoPOBXYCHwWetKdfeFaHvu1fabuGKTrZdv9H5s2SvmT7ZEkvbq2q3f0UeJztuwAkHQJcCJwEfAFoNfRtr2nz9R+EXbYvbLuIMTwU2J8mKx7e134P8PxWKhqD7f9su4YpOMn28ZKuBbD93TJhdY+b1aHfT9IxNMs5P6zXZvuS9ioaaH9JJ9m+GkDSiTR/cAC72itrN4f3Ar+4GzjS9nck/aStokaTNA/4U3b/d+/aJ7x/kvQHwMeBe3uNtr/TXkn31/B54POSLp4poSppCfBW4HE0b1pzgB/YnttqYSP9pCw1b7j//+pPh/HCVYS+pNfRdEEcBfwzzXr+XwS6FvovA94taX+afsh7gJdJ2g/461YrG+nfJF1G0y0B8JvAF0qd32utqt29H/gQcCrw+8AKYEerFQ22oty+tq/NwKNaqGUESX9v+1XA2yQNWhzxecOvakJvo+ka/TCwGDgD+B+tVrS782ne5A+WtJrmU9NfDOOFZ/WJ3J5yguc44Frbx5XuiHfa7kof+QiSHkHzb/O9tmsZpJyE+k3gZJo3py8CH3XH/jNJ2mT7BElf753Ek/R5209ru7aZQtIJtjdJGvg7K58EOqW3YuWof/crbT+l7dr6SXoscArN39AG2zcO43WrONIHfmT7p5J2SZpL0x3R+lHUaJL2oQnTw4G9m2wF2+e2WNZuSrh/pHx1Wa+rabukU2lWdF3QYj0DSXoIcBbwy6Xpc8A7bLfeVWZ7U7m9P9wlHQgstP311gob3w9L//h1kt4MbAf2a7mmQe4C/o0mh/eVdPwwRhTWEvobJR1AMzpiE/DfwB4fGvUgrKMZ+riJvr7drilDNv8GOJjmKKU3JK5LfaYAbyyfml5D08c7F/ijdksa6ELgITQjOgBeUtpe1lpFo0j6HPA8msy4DthRPjW9erz9WvISmjlIL6f5915IczDVGZL+Cvgd4D94YJjpUEYUVtG900/S4cDcLh6lSLre9jFt1zERSVuB5w7r4+hsJ+lrto+bqK1Nkq61/URJL6M5yn9df/dJTI2km4DHl8vGDlU1M3IlzZf0FOAw4ABJvzzRPi24UtLj2y5iEu6aCYEvaU35hNfbPlDSu1ssaSz3SXp0b0PSo+jYeH2a7sZDgRcAl7VdzHgknSzpCkn/LumW3lfbdY1yPXBAGy9cRfeOpL8Bfgu4gQf+mEwzprxLngr8jqRbabp3OjeTsNgo6UPAJxg5xPBjrVU02LH9J8PLWOgntljPWF4LfLYEk2hm5r603ZJ2cy7NdTC+aPsr5Y3p5pZrGsu7aLp1NtG9N8+evwaulXQ9I/+G9vhoqCq6d8pHqWNtd7afHEDSIwe1d218tKT3DGi27d8dejHjkPQ14Om2v1u2DwI+b7tzn6bKSfzH0IT+N7r+f7XLJF1t+6S26xiPpC3AO4DN9I3PH8ZoqCqO9IFbaE6UdfIPSdJc2/fQTMXuPNtdOwody1tousx6o4xOB1a3WM8Ikp5h+18HrGX0aEmd+uRUJg/9HmVkWa+9a2/0xWcl/S3wMUYeRXdpra1v2T6/jReuJfR/SDN8awMj/xN0ZeGtD9CsG7KJpttJfd/rxCQdAEl/YvvNYy1o1qHfJ9DMuJa0kWZEhIDfsH1Dy2X1exrwrwxeU8k0odUV62iGF/4L3e0y6ekd5S/ua+vaWlubJP01zTXCh/rGVEv3zopB7TNwjZZWSXqu7X/q+u+z98mpdOfspgvLG/STdITtWydqa5Ok62w/oe06ZgtJnx3Q7GEsEVJF6AOUyRpHls2bujDxZZDyUf+pNEcm/2b7E+1WNPNIusz2c8oJ8f7/4L0T45345NSjwcsWb7J9Qls1jSbpjcCVtv+57VomUuZmvI4HJrt9HjjX9vfbq6o7qgh9SU8H1gC30fzhLwRW2O7U6B1JF9CsEfLB0vRbwH/YPru9qnanZv33P2b3/t0ufXzuvDIN/2jgzYxcd2cu8FrbR7dS2ACSdtLMar2XZqZzVyfkIemjNEMie588XwIcZ3vQdSBa0eYbUy2hvwl4ke2byvaRwAe7dCQF95/RP6a3ho2kvWguTNKZP364f1TM2xk1JK43Zb9tmiEXz5G0DDiNZqbr+r5v7QQutX1lG3XNdIO6orrWPdXmG1MtJ3If0gt8ANv/XtY76ZqbaCaP9YZoLgQ6N3OYbq//Dg9cPOdhNCfzvkZzZHoscDVN91nrbK8D1kl6su2r2q5nImXNnUWMXKa6U5+Wix9JeqrtL8L9l3X8Ucs1jfZo2/1LQ7xB0nXDeOFaQn+jpHfxwMU9XkxzlNo1Pw/cKKm3LtCTgKskrYdOLWPb2fXf4YGL50i6FFhpe3PZPoamW6prrpV0Nk1XT3+gdmY4ZFl+4ZU0C9ZdBywBrqJbI2J6zgLWlC4UAd+hWeemS1p7Y6qle2cf4GyaIzzRzMS9oGsTYDTG8rU9XVnGtpwgHa2LJ0g7/zEfQNKHgW8AL6KZ+frbwI22X9lqYX3ULE/+JODLtp9Qzke8wfZvtVzamNSsqEuZA9Mpkp5A07XziNL0XZrzjHv8k30Vod+vDONb0MUF1wAk/SJwIs2ok6/Y/mbLJc1Ykj4I/AB4H83v88XA/rZf2Gpho/QtZvZ128eWrsdPd+nEuKSv2H5S6YI4yfa9XXsDlfRi2+/TyGv53s/23w27prGUA9HnA4+mWYPn+zQHTnt8GfUqunc0Q5aFLR+h/5Jmwo6At6q50HgnFgkbZwYp0Mm1d15K81G/d8T8BZoli7umN3z4e6UL6ps0I6O6ZFtZvO4TwBWSvktzfYIu6a2Z//AB3+va0e06mqvMfRX4r2G+cBVH+pohy8KWNYKeYvvbZfvnacZGP6bdyhqSXm/79WXtnd7M4ftvu9QH3SNpX+Cw/hP5XVP+X34UeDxwMc11kf/S9tvbrGsspRvyEcCn3MLSwBORdLLtL03U1ia1uIx6FUf6jFwW9s/bLmYc2xi5/s5O4I6WahlkZ/nofD0jl4vo5JGDpOcBf0tzcewjSj/quR06IQ6A7XeWu1+gI0tu9CtDh7/eC6munFsax1uB0cN2B7W16UpJj+8NMhimWkL/DcyMZWH/C7ha0jqaIF0GXNPro+xAn+T+5fYxNCf11tEE/3Pp3jLV0Ex+OZHm8oPYvk7NRXQ6RdKbgDe7LANdhka+xvZQLpQ9ETeXGv2apMNs3952PWOR9GTgKcC8Uf36c4E57VQ1Ujkhbprsfama5bSHuoz6rA99SXNounTu/2XavoWOXT6t+I/y1bOu3A7qoxw6228AkPQZ4HjbO8v264EPt1jaWHbZ/r6kiR/Zrmfb/rPehpt1/38N6EToF4cCW8pw4h/0Gjv2qemhNAcmezPyb+YempOmXfCctguY9aFv+77yMf//tF3LRPpC9eHNpv+75ZLGchjQ35f7Y7p34hHgekkvAuZIWgT8IdDFWa5zJO3TG0JczkPs03JNo+3PyMASzXWSO6N0O31e0sXu2DUoerpQ16wP/eJKSW8DPsTIo5ROTMfvKSM33gscVLa/BZxhe0urhe3uvTTdTh+n+aj66zwwnbxLXkFzDudemuWrPw38VasVDfY+YEPfCfLfpXu/z71H9+WXN6cu+qGa9fRHT3brzBDYNtUyeqe1ZUynQtKVwJ/b/mzZfjrwJttPabOuQcr6Nv+zbH7B9rVt1jOIpMU0oX84DxzgDKXfdKokPRs4heYI+jO2P91ySQBIOgv4A5oTzP1djw8HvmT7xa0UNo7S/fghmtnXvw+sAHbY/tNWC+uIKkJ/ppD0NdvHTdQWk1OGwP4xzWij/kvStf4Re6YoSxkcSHNN11V939rZlWU3RlNZlrp/WHaZlzPujPdaVNG9I+kQ4E3AL9l+tqSjgCfbflfLpY12i6T/zcg1gjpzIY0ZaIftf2q7iImoWba4d/T1UJpLe/7AHVi22M1Sv98HOjWLeQK9yW7bJZ1KM4lsQYv1dEoVR/qSLgfeQ9N1cpykvYFr3bELZJehem9g5BpBr3e5sHdMjaRTaMJq9GUyuzZzeARJpwEn9o/oicmT9ByaSzsupBmfP5dmnaD14+5YiVpCv7duyLW2n1jaOrVuSEw/Se8DHgts4YHunU7OHB5N0pdtL2m7jph9qujeAX5QljToXZxkCc1H1k5Rrkg13Y7r2qe5QUatZbQXzTUAZv/R2B5S/o4uBA6xfYykY4Hn2X5jy6V1Qi2h/2qaKxM9StKXgHl0Z7JGvw/TXJHqnfRdkSoetC9LOsr2DW0XMoHn9t3fRXNZz2XtlDIr/CPN5SffAWD765I+ACT0qSf0b6C54McPadaz+QTw720WNIauX5FqpnkqsKKs/z/Uqe5TYfulbdcwy/yc7WtGzcTe1VYxXVNL6F9CMxX7TWX7hTQjZE5vraI+ZY1/6PgVqWagpW0XMB5Jb2WcbhzbfzjEcmaTb0l6NA905z4f2N5uSd1Ry4ncTo9/L0ei/atWQl8YuGNXpIrpIWlFuXsycBTNhCJoDkY22f6jVgqb4cqCihfRLL72XZphz7+d+RmNWo70r5W0xPaXASSdBHRmbW3bRwBIegHNGuX3lPH6x9PNZQNiGtheAyDpd4Bfsf2Tsv124DMtljZjlQUWz7L9TEn7AXv1FgaMxl5tFzAkJ9Gsv3ObpNtoLuj8NEmbJXXpsol/UQL/qcD/ormgRvr4Z79fYuSqkPuXtpgi2/cBJ5T7P0jg766WI/1O9+326Y3YORV4u+11ZdnimN3OA76q5rKeAE8DXt9aNTPftZLW04yG619gsdOT8oalij79mULSZTQXUnkmzdHKj4BrunLuIfYMNcNMXgK8iibsrwN+0fY17VU1c5XVSkebEZPyhiGh3yGSfo7mU8lm2zeXSzw+3nb6d2cxSRfSzBh+hu3HleU4PmP7SS2XFrNQLX36M4LtH9r+mO2by/b2BH4VTrJ9NvD/oLlyFs3Ca/EgSFoj6YC+7QMlvbvFkjoloR/Rvp+UUSe9ceXz6FsKOqbs2N71huH+N9EntldOtyT0I9p3Ps2EvIMlrQa+yAMTCWPq9ipdZMD9kx9rGbQyofwiIlpm+/2SNvHAlbNOs31jy2XNZG+hGaL9EZpPTy8AVrdbUnfkRG5EzDrlQknPoHkT3TADFt0bmnTvRMRsdBDN1cfeCuyQdETbBXVFjvQjYlaR9DqaaxI8xvaRkn4J+LDtk1surRNypB8Rs82vA8+jzMa1fScjl7moWkI/ImabH7vpwugNgd2v5Xo6JaEfEbNGWdLiMknvAA6Q9HvAv9BcTStIn35EzDKSvgr8KfCrNKN3Pm37inar6o6M04+I2eYq4Hu2X9t2IV2UI/2ImFUk3QAcCfwnI5dW7tS1kduS0I+IWUXSIwe153KJjYR+RERFMnonIqIiCf2IiIok9CMiKpLQj4ioSEI/IqIi/x9o+bHsRjKWRwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "places_data['purpose'].value_counts().plot(kind='bar')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "b9bf781a",
   "metadata": {},
   "outputs": [],
   "source": [
    "CategoriesToSelect = ['education','medical','shopping', 'personal', 'dining', 'transit'] # select the required categories for the ML Model \n",
    "places_data = places_data[places_data['purpose'].isin(CategoriesToSelect)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "dcdd8e3d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(15290, 14)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "places_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "f219c2f2",
   "metadata": {},
   "outputs": [],
   "source": [
    "places_data = places_data[['clean_name','purpose']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "fafb5dda",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "places_data['purpose'].value_counts().plot(kind='bar')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "01557c6b",
   "metadata": {},
   "outputs": [],
   "source": [
    "places_data = places_data.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "8a619424",
   "metadata": {},
   "outputs": [],
   "source": [
    "vectorizer =  CountVectorizer()\n",
    "\n",
    "bow_matrix = vectorizer.fit_transform(places_data['clean_name'])\n",
    "\n",
    "word_features = vectorizer.get_feature_names()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "5d48dc88",
   "metadata": {},
   "outputs": [],
   "source": [
    "bow_table = pd.DataFrame(bow_matrix.toarray() , columns = word_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "64c2da02",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(15290, 11276)"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bow_table.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "2181e0be",
   "metadata": {},
   "outputs": [],
   "source": [
    "## trying different feature extraction techniques "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f3d091d1",
   "metadata": {},
   "source": [
    "tsvd = TruncatedSVD(n_components=10)\n",
    "reduced_features = tsvd.fit_transform(bow_matrix)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f92225b3",
   "metadata": {},
   "source": [
    "pca = PCA(n_components=10, whiten=True)\n",
    "reduced_features = pca.fit_transform(bow_table)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "54517401",
   "metadata": {},
   "outputs": [],
   "source": [
    "kpca = KernelPCA(n_components=0.99, whiten=True)\n",
    "reducted_features = kpca.fit_transform(bow_table)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9602a289",
   "metadata": {},
   "source": [
    "lda = LinearDiscriminantAnalysis(n_components=1)\n",
    "reducted_features = lda.fit(bow_table, places_data['purpose']).transform(bow_table)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6dfb41ba",
   "metadata": {},
   "source": [
    "nmf = NMF(n_components=1000, random_state=1)\n",
    "reduced_features = nmf.fit_transform(bow_matrix)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0054d490",
   "metadata": {},
   "outputs": [],
   "source": [
    "## trying different feature selection techniques"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "42157f4e",
   "metadata": {},
   "source": [
    "thresholder = VarianceThreshold(threshold=(.75 * (1 - .75)))\n",
    "reduced_features = thresholder.fit_transform(bow_matrix)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "20fe7513",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<15290x11276 sparse matrix of type '<class 'numpy.int64'>'\n",
       "\twith 45532 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reduced_features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "8816ca82",
   "metadata": {},
   "outputs": [],
   "source": [
    "reduced_names = pd.DataFrame(reduced_features, columns = np.arange(1,11,1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "f1364616",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(15290, 10)"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reduced_names.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "e8da4b0c",
   "metadata": {},
   "outputs": [],
   "source": [
    "bow_table.drop([col for col, val in bow_table.sum().iteritems() if val == 1], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "16b3147e",
   "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>aa</th>\n",
       "      <th>aaa</th>\n",
       "      <th>aadare</th>\n",
       "      <th>aalindaya</th>\n",
       "      <th>aaliyah</th>\n",
       "      <th>aari</th>\n",
       "      <th>aashkii</th>\n",
       "      <th>aaysha</th>\n",
       "      <th>abacus</th>\n",
       "      <th>abans</th>\n",
       "      <th>...</th>\n",
       "      <th>zen</th>\n",
       "      <th>zenith</th>\n",
       "      <th>zero</th>\n",
       "      <th>zion</th>\n",
       "      <th>zodiac</th>\n",
       "      <th>zone</th>\n",
       "      <th>zoo</th>\n",
       "      <th>zoysa</th>\n",
       "      <th>zuls</th>\n",
       "      <th>zumba</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15285</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15286</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15287</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15288</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15289</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>15290 rows × 3853 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       aa  aaa  aadare  aalindaya  aaliyah  aari  aashkii  aaysha  abacus  \\\n",
       "0       0    0       0          0        0     0        0       0       0   \n",
       "1       0    0       0          0        0     0        0       0       1   \n",
       "2       0    0       0          0        0     0        0       0       0   \n",
       "3       0    0       0          0        0     0        0       0       0   \n",
       "4       0    0       0          0        0     0        0       0       0   \n",
       "...    ..  ...     ...        ...      ...   ...      ...     ...     ...   \n",
       "15285   0    0       0          0        0     0        0       0       0   \n",
       "15286   0    0       0          0        0     0        0       0       0   \n",
       "15287   0    0       0          0        0     0        0       0       0   \n",
       "15288   0    0       0          0        0     0        0       0       0   \n",
       "15289   0    0       0          0        0     0        0       0       0   \n",
       "\n",
       "       abans  ...  zen  zenith  zero  zion  zodiac  zone  zoo  zoysa  zuls  \\\n",
       "0          0  ...    0       0     0     0       0     0    0      0     0   \n",
       "1          0  ...    0       0     0     0       0     0    0      0     0   \n",
       "2          0  ...    0       0     0     0       0     0    0      0     0   \n",
       "3          0  ...    0       0     0     0       0     0    0      0     0   \n",
       "4          0  ...    0       0     0     0       0     0    0      0     0   \n",
       "...      ...  ...  ...     ...   ...   ...     ...   ...  ...    ...   ...   \n",
       "15285      0  ...    0       0     0     0       0     0    0      0     0   \n",
       "15286      0  ...    0       0     0     0       0     0    0      0     0   \n",
       "15287      0  ...    0       0     0     0       0     0    0      0     0   \n",
       "15288      0  ...    0       0     0     0       0     0    0      0     0   \n",
       "15289      0  ...    0       0     0     0       0     0    0      0     0   \n",
       "\n",
       "       zumba  \n",
       "0          0  \n",
       "1          0  \n",
       "2          0  \n",
       "3          0  \n",
       "4          0  \n",
       "...      ...  \n",
       "15285      0  \n",
       "15286      0  \n",
       "15287      0  \n",
       "15288      0  \n",
       "15289      0  \n",
       "\n",
       "[15290 rows x 3853 columns]"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bow_table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "c088eb45",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train,X_test,y_train,y_test = train_test_split(bow_table,places_data['purpose'],test_size=0.2, random_state=32)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "de176e28",
   "metadata": {},
   "source": [
    "### Naive Bayes Model "
   ]
  },
  {
   "cell_type": "raw",
   "id": "7e9ec5d7",
   "metadata": {},
   "source": [
    "nb = Pipeline([('vec', CountVectorizer()) ,\n",
    "              ('clf', MultinomialNB(fit_prior=True))])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "1ba6df88",
   "metadata": {},
   "outputs": [],
   "source": [
    "nb = MultinomialNB(fit_prior=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "1a969190",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MultinomialNB()"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nb.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "86cf5435",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = nb.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "id": "a1028999",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7354480052321779"
      ]
     },
     "execution_count": 87,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nb.score(X_test,y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "id": "fcf0c81c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "      dining       0.66      0.75      0.70       398\n",
      "   education       0.75      0.85      0.80       210\n",
      "     medical       0.70      0.83      0.76       216\n",
      "    personal       0.82      0.70      0.75      1368\n",
      "    shopping       0.67      0.72      0.69       807\n",
      "     transit       0.86      1.00      0.92        59\n",
      "\n",
      "    accuracy                           0.74      3058\n",
      "   macro avg       0.74      0.81      0.77      3058\n",
      "weighted avg       0.74      0.74      0.74      3058\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_pred,y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f2dc9991",
   "metadata": {},
   "source": [
    "### SVM Classifier from SGDC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "0b0c3bb7",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import SGDClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bbcd0d34",
   "metadata": {},
   "source": [
    "sgd = Pipeline([('vect', CountVectorizer()),\n",
    "                ('clf', SGDClassifier(loss='modified_huber', penalty='l2',alpha=1e-3, random_state=32, \n",
    "                                      max_iter=100, tol=None, learning_rate = 'optimal', class_weight = 'balanced'))])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "id": "aa63d26b",
   "metadata": {},
   "outputs": [],
   "source": [
    "paramters = {'loss':np.array(['modified_huber', 'hinge', 'log',  'squared_hinge' , 'perceptron'])}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "138da527",
   "metadata": {},
   "outputs": [],
   "source": [
    "sgd =  SGDClassifier(loss = 'modified_huber', penalty='l2',alpha=1e-3, random_state=32, \n",
    "                                      max_iter=100, tol=None, learning_rate = 'optimal', class_weight = 'balanced')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "604b40b2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SGDClassifier(alpha=0.001, class_weight='balanced', loss='modified_huber',\n",
       "              max_iter=100, random_state=32, tol=None)"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sgd.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "6bc15f4a",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = sgd.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "e5c89103",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy 0.7586657946370177\n"
     ]
    }
   ],
   "source": [
    "print('accuracy %s' % accuracy_score(y_pred, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "23601c2a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "      dining       0.76      0.70      0.73       457\n",
      "   education       0.82      0.77      0.79       240\n",
      "     medical       0.81      0.79      0.80       256\n",
      "    personal       0.84      0.73      0.78      1164\n",
      "    shopping       0.65      0.81      0.72       872\n",
      "     transit       0.79      0.93      0.85        69\n",
      "\n",
      "    accuracy                           0.76      3058\n",
      "   macro avg       0.78      0.79      0.78      3058\n",
      "weighted avg       0.77      0.76      0.76      3058\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "id": "d3fe5cbc",
   "metadata": {},
   "outputs": [],
   "source": [
    "sgdlr =  SGDClassifier(loss = 'modified_huber', penalty='l1',alpha=1e-3, random_state=32, \n",
    "                                      max_iter=100, tol=None, learning_rate = 'optimal', class_weight = 'balanced')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "id": "c8703e66",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SGDClassifier(alpha=0.001, class_weight='balanced', loss='modified_huber',\n",
       "              max_iter=100, penalty='l1', random_state=32, tol=None)"
      ]
     },
     "execution_count": 168,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sgdlr.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "id": "7bf5cbaa",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = sgdlr.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "id": "2b41dbba",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy 0.6756049705689994\n"
     ]
    }
   ],
   "source": [
    "print('accuracy %s' % accuracy_score(y_pred, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "id": "1d923de8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "      dining       0.75      0.58      0.65       457\n",
      "   education       0.81      0.70      0.75       240\n",
      "     medical       0.76      0.71      0.73       256\n",
      "    personal       0.85      0.56      0.68      1164\n",
      "    shopping       0.53      0.84      0.65       872\n",
      "     transit       0.62      0.93      0.74        69\n",
      "\n",
      "    accuracy                           0.68      3058\n",
      "   macro avg       0.72      0.72      0.70      3058\n",
      "weighted avg       0.73      0.68      0.68      3058\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a82b94ef",
   "metadata": {},
   "source": [
    "### Logistics Regression Classifier "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "25732128",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "99cd25df",
   "metadata": {},
   "source": [
    "pipeline = Pipeline([('vect', CountVectorizer()),\n",
    "                ('clf', LogisticRegression())])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "c8f3c019",
   "metadata": {},
   "outputs": [],
   "source": [
    "clf = LogisticRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "id": "93e8a5c2",
   "metadata": {},
   "outputs": [],
   "source": [
    "parameters = {'solver':np.array(['newton-cg', 'lbfgs', 'liblinear', 'sag', 'saga'])}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "id": "006b0278",
   "metadata": {},
   "outputs": [],
   "source": [
    "logreg = GridSearchCV(clf, param_grid = parameters)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b6202b2e",
   "metadata": {},
   "source": [
    "logreg = LogisticRegression(max_iter=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "8d118410",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\HP\\anaconda3\\envs\\ML1\\lib\\site-packages\\sklearn\\linear_model\\_logistic.py:763: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "LogisticRegression()"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "b20ba055",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = clf.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "3a74b1ca",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy 0.7586657946370177\n"
     ]
    }
   ],
   "source": [
    "print('accuracy %s' % accuracy_score(y_pred, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "id": "12b8a9a2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'solver': 'newton-cg'}"
      ]
     },
     "execution_count": 158,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "logreg.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f54e23fd",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(classification_report(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e56e8944",
   "metadata": {},
   "source": [
    "### KNN "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "64cc7e1e",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.neighbors import KNeighborsClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e78ad18b",
   "metadata": {},
   "source": [
    "pipeline = Pipeline([('vec', CountVectorizer()),('knn',KNeighborsClassifier())])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "f17be6b2",
   "metadata": {},
   "outputs": [],
   "source": [
    "knn = KNeighborsClassifier()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "id": "84b63036",
   "metadata": {},
   "outputs": [],
   "source": [
    "parameters = {'n_neighbors':np.arange(1,50)}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "id": "7b5d9cb0",
   "metadata": {},
   "outputs": [],
   "source": [
    "knn = GridSearchCV(knn, param_grid = parameters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "f23b0f88",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "KNeighborsClassifier()"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knn.fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "8c7e0093",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = knn.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "dbaae21b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy 0.6262262916939176\n"
     ]
    }
   ],
   "source": [
    "print('accuracy %s' % accuracy_score(y_pred, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "7e6d6d92",
   "metadata": {},
   "outputs": [
    {
     "ename": "AttributeError",
     "evalue": "'KNeighborsClassifier' object has no attribute 'get_parameters'",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mAttributeError\u001b[0m                            Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-59-e2e941743b88>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mknn\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mget_parameters\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;31mAttributeError\u001b[0m: 'KNeighborsClassifier' object has no attribute 'get_parameters'"
     ]
    }
   ],
   "source": [
    "knn.params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "id": "e6f7000f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "      dining       0.63      0.53      0.58       457\n",
      "   education       0.85      0.62      0.72       240\n",
      "     medical       0.68      0.61      0.64       256\n",
      "    personal       0.60      0.73      0.66      1164\n",
      "    shopping       0.60      0.53      0.56       872\n",
      "     transit       0.72      0.93      0.81        69\n",
      "\n",
      "    accuracy                           0.63      3058\n",
      "   macro avg       0.68      0.66      0.66      3058\n",
      "weighted avg       0.63      0.63      0.62      3058\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8800a75f",
   "metadata": {},
   "source": [
    "### Random Forest "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "034d1724",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3373382d",
   "metadata": {},
   "source": [
    "pipeline = Pipeline([('vec', CountVectorizer()), ('clf',RandomForestClassifier(n_estimators=100,))])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "id": "fe15f1d5",
   "metadata": {},
   "outputs": [],
   "source": [
    "parameters = {'n_estimators':np.array([20,50,80,100,150,200])}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "5eb31bfe",
   "metadata": {},
   "outputs": [],
   "source": [
    "clf = RandomForestClassifier(n_estimators= 150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "id": "28918376",
   "metadata": {},
   "outputs": [],
   "source": [
    "rfc = GridSearchCV(clf, param_grid = parameters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "673a8afc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(n_estimators=150)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "id": "ec4bd67a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'n_estimators': 150}"
      ]
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rfc.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "afe8900a",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = clf.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "91567293",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy 0.7289077828646174\n"
     ]
    }
   ],
   "source": [
    "print('accuracy %s' % accuracy_score(y_pred, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "ae6ef0bb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "      dining       0.75      0.64      0.69       457\n",
      "   education       0.85      0.69      0.76       240\n",
      "     medical       0.80      0.71      0.75       256\n",
      "    personal       0.79      0.73      0.76      1164\n",
      "    shopping       0.62      0.78      0.69       872\n",
      "     transit       0.84      0.90      0.87        69\n",
      "\n",
      "    accuracy                           0.73      3058\n",
      "   macro avg       0.77      0.74      0.75      3058\n",
      "weighted avg       0.74      0.73      0.73      3058\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d636b5a2",
   "metadata": {},
   "source": [
    "###  keras NN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 376,
   "id": "aaee674e",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import LabelBinarizer, LabelEncoder\n",
    "from sklearn.metrics import confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 370,
   "id": "1f98cdad",
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Activation, Dropout\n",
    "from keras.preprocessing import text, sequence\n",
    "from keras import utils"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 372,
   "id": "d4cb91c2",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_size = int(len(places_data) * .7)\n",
    "train_posts = places_data['clean_name'][:train_size]\n",
    "train_tags = places_data['purpose'][:train_size]\n",
    "\n",
    "test_posts = places_data['clean_name'][train_size:]\n",
    "test_tags = places_data['purpose'][train_size:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 396,
   "id": "3d0fc1e4",
   "metadata": {},
   "outputs": [],
   "source": [
    "max_words = 10000\n",
    "tokenize = text.Tokenizer(num_words=max_words, char_level=False)\n",
    "tokenize.fit_on_texts(train_posts) # only fit on train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 397,
   "id": "f5a5b0db",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train = tokenize.texts_to_matrix(train_posts)\n",
    "x_test = tokenize.texts_to_matrix(test_posts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 398,
   "id": "2e3c712f",
   "metadata": {},
   "outputs": [],
   "source": [
    "encoder = LabelEncoder()\n",
    "encoder.fit(train_tags)\n",
    "y_train = encoder.transform(train_tags)\n",
    "y_test = encoder.transform(test_tags)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 399,
   "id": "5be7b18e",
   "metadata": {},
   "outputs": [],
   "source": [
    "num_classes = np.max(y_train) + 1\n",
    "y_train = utils.to_categorical(y_train, num_classes)\n",
    "y_test = utils.to_categorical(y_test, num_classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 400,
   "id": "8b07946a",
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 32\n",
    "epochs = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 401,
   "id": "83f91031",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/2\n",
      "301/301 [==============================] - 6s 21ms/step - loss: 1.0775 - accuracy: 0.5986 - val_loss: 0.7615 - val_accuracy: 0.7292\n",
      "Epoch 2/2\n",
      "301/301 [==============================] - 6s 21ms/step - loss: 0.5080 - accuracy: 0.8384 - val_loss: 0.7050 - val_accuracy: 0.7376\n"
     ]
    }
   ],
   "source": [
    "# Build the model\n",
    "model = Sequential()\n",
    "model.add(Dense(512, input_shape=(max_words,)))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(num_classes))\n",
    "model.add(Activation('softmax'))\n",
    "\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer='adam',\n",
    "              metrics=['accuracy'])\n",
    "              \n",
    "history = model.fit(x_train, y_train,\n",
    "                    batch_size=batch_size,\n",
    "                    epochs=epochs,\n",
    "                    verbose=1,\n",
    "                    validation_split=0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 402,
   "id": "ac1ab816",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "144/144 [==============================] - 1s 5ms/step - loss: 0.8713 - accuracy: 0.6787\n",
      "Test accuracy: 0.678657054901123\n"
     ]
    }
   ],
   "source": [
    "score = model.evaluate(x_test, y_test,\n",
    "                       batch_size=batch_size, verbose=1)\n",
    "print('Test accuracy:', score[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "056218fb",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cff0d7d7",
   "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"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}