airbnb_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import pickle\n",
    "import herepy\n",
    "from tqdm.notebook import tqdm\n",
    "from collections import defaultdict\n",
    "import re\n",
    "import requests\n",
    "import urllib\n",
    "import xmltodict\n",
    "import json\n",
    "import numpy as np\n",
    "#packages for R in python\n",
    "import rpy2.robjects as ro\n",
    "from rpy2.robjects import r, pandas2ri\n",
    "from rpy2.robjects.packages import importr\n",
    "#CPD\n",
    "from pgmpy.models import BayesianModel\n",
    "import pgmpy\n",
    "# from pgmpy.estimators import ConstraintBasedEstimator\n",
    "#bnlearn\n",
    "import bnlearn\n",
    "import torch\n",
    "import pyro\n",
    "import pyro.distributions as dist\n",
    "from statistics import mean\n",
    "from pyro.infer import Importance, EmpiricalMarginal\n",
    "import matplotlib.pyplot as plt\n",
    "from collections import defaultdict\n",
    "from IPython.display import Image\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "listings_full = pd.read_csv('data/listings_full.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import geopandas as gpd\n",
    "import geopy\n",
    "from geopy.geocoders import Nominatim\n",
    "from geopy.extra.rate_limiter import RateLimiter\n",
    "\n",
    "def herePy(coordinate1,coordinate2):\n",
    "    coordinates = coordinate1+','+coordinate2\n",
    "    locator = Nominatim(user_agent=\"myGeocoder\")\n",
    "    address = locator.reverse(coordinates)\n",
    "    address = address.raw\n",
    "    return pd.Series([address.get('display_name',None),\n",
    "                      address['address'].get('house_number',address['address'].get('building',None)),\n",
    "                      address['address'].get('road',None),\n",
    "                      address['address'].get('country',None),\n",
    "                      address['address'].get('postcode',None)],\n",
    "                     index=['Address','House','Street','Country',\n",
    "                            'PostalCode'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/abdulrehman/opt/anaconda3/envs/Airbnb_Analysis/lib/python3.8/site-packages/tqdm/std.py:699: FutureWarning: The Panel class is removed from pandas. Accessing it from the top-level namespace will also be removed in the next version\n",
      "  from pandas import Panel\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5d14159fd7af4e63a09649748288393e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(HTML(value=''), FloatProgress(value=0.0, max=200.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "tqdm.pandas()\n",
    "\n",
    "listings_full[['Address','HouseNumber','Street','Country','PostalCode']] = listings_full.progress_apply(lambda row: herePy(str(row['latitude']), str(row['longitude'])) , axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# listings_full['Street'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "listings_full.to_csv('data/listings_full2.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "listings_subset = listings_full[['id','Zestimate','bedrooms','bathrooms','beds','accommodates','amenities_score',\n",
    "               'bikeScore','walkScore','transitScore','neighbourhood_cleansed','property_type',\n",
    "               'publicSchoolScores','closenessToDowntown','price','minimum_nights_avg_ntm','roi','Street']]\n",
    "\n",
    "pd.options.mode.chained_assignment = None\n",
    "listings_subset['property_type'] = listings_subset['property_type'].astype('category')\n",
    "listings_subset['bathrooms'] = listings_subset['bathrooms'].astype('category')\n",
    "listings_subset['neighbourhood'] = listings_subset['neighbourhood_cleansed'].astype('category')\n",
    "listings_subset['street'] = listings_subset['Street'].astype('category')\n",
    "del listings_subset['neighbourhood_cleansed']\n",
    "del listings_subset['Street']\n",
    "\n",
    "listings_subset['roi'] = listings_subset['roi'].astype('float')\n",
    "listings_subset['Zestimate'] = listings_subset['Zestimate'].astype('float')\n",
    "listings_subset['bedrooms'] = listings_subset['bedrooms'].astype('float')\n",
    "listings_subset['beds'] = listings_subset['beds'].astype('float')\n",
    "listings_subset['accommodates'] = listings_subset['accommodates'].astype('float')\n",
    "listings_subset['amenities_score'] = listings_subset['amenities_score'].astype('float')\n",
    "listings_subset['publicSchoolScores'] = listings_subset['publicSchoolScores'].astype('float')\n",
    "listings_subset['closenessToDowntown'] = listings_subset['closenessToDowntown'].astype('float')\n",
    "listings_subset['price'] = listings_subset['price'].astype('float')\n",
    "listings_subset['minimum_nights_avg_ntm'] = listings_subset['minimum_nights_avg_ntm'].astype('float')\n",
    "pd.options.mode.chained_assignment = 'warn'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "listing_cat = listings_subset[['bathrooms','bikeScore','walkScore','transitScore','property_type','neighbourhood','street']]\n",
    "listings_n_cat = listings_subset[['roi','Zestimate','bedrooms','beds','accommodates','amenities_score','publicSchoolScores','closenessToDowntown','price','minimum_nights_avg_ntm']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def rDiscretize(data):\n",
    "    utils = importr('utils')\n",
    "    utils.install_packages('bnlearn')\n",
    "    # utils.install_packages('base')\n",
    "    bnlearnR = importr('bnlearn')\n",
    "    bnDiscretize = bnlearnR.discretize\n",
    "    # Import Functions\n",
    "    pandas2ri.activate()\n",
    "    r_dt = ro.conversion.py2rpy(data) #convert pandas to RDataframe\n",
    "    r_dt2 = bnDiscretize(r_dt,method= \"hartemink\", breaks = 3, ibreaks=3)\n",
    "    pd_dt = ro.conversion.rpy2py(r_dt2)\n",
    "    pd_dt.reset_index(drop=True)\n",
    "    return pd_dt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- Please select a CRAN mirror for use in this session ---\n",
      "Secure CRAN mirrors \n",
      "\n",
      " 1: 0-Cloud [https]\n",
      " 2: Australia (Canberra) [https]\n",
      " 3: Australia (Melbourne 1) [https]\n",
      " 4: Australia (Melbourne 2) [https]\n",
      " 5: Australia (Perth) [https]\n",
      " 6: Austria [https]\n",
      " 7: Belgium (Ghent) [https]\n",
      " 8: Brazil (BA) [https]\n",
      " 9: Brazil (PR) [https]\n",
      "10: Brazil (RJ) [https]\n",
      "11: Brazil (SP 1) [https]\n",
      "12: Brazil (SP 2) [https]\n",
      "13: Bulgaria [https]\n",
      "14: Canada (MB) [https]\n",
      "15: Chile (Santiago) [https]\n",
      "16: China (Beijing 2) [https]\n",
      "17: China (Hefei) [https]\n",
      "18: China (Hong Kong) [https]\n",
      "19: China (Guangzhou) [https]\n",
      "20: China (Lanzhou) [https]\n",
      "21: China (Nanjing) [https]\n",
      "22: China (Shanghai 1) [https]\n",
      "23: China (Shanghai 2) [https]\n",
      "24: Costa Rica [https]\n",
      "25: Denmark [https]\n",
      "26: East Asia [https]\n",
      "27: Ecuador (Cuenca) [https]\n",
      "28: Ecuador (Quito) [https]\n",
      "29: Estonia [https]\n",
      "30: France (Lyon 1) [https]\n",
      "31: France (Lyon 2) [https]\n",
      "32: France (Marseille) [https]\n",
      "33: France (Montpellier) [https]\n",
      "34: Germany (Erlangen) [https]\n",
      "35: Germany (Leipzig) [https]\n",
      "36: Germany (Göttingen) [https]\n",
      "37: Germany (Münster) [https]\n",
      "38: Germany (Regensburg) [https]\n",
      "39: Greece [https]\n",
      "40: Hungary [https]\n",
      "41: Iceland [https]\n",
      "42: India [https]\n",
      "43: Indonesia (Jakarta) [https]\n",
      "44: Iran [https]\n",
      "45: Italy (Milano) [https]\n",
      "46: Italy (Padua) [https]\n",
      "47: Japan (Tokyo) [https]\n",
      "48: Korea (Gyeongsan-si) [https]\n",
      "49: Korea (Seoul 1) [https]\n",
      "50: Korea (Ulsan) [https]\n",
      "51: Malaysia [https]\n",
      "52: Mexico (Mexico City) [https]\n",
      "53: Morocco [https]\n",
      "54: Netherlands [https]\n",
      "55: Norway [https]\n",
      "56: Philippines [https]\n",
      "57: Russia (Moscow) [https]\n",
      "58: South Africa (Johannesburg) [https]\n",
      "59: Spain (A Coruña) [https]\n",
      "60: Spain (Madrid) [https]\n",
      "61: Sweden (Borås) [https]\n",
      "62: Sweden (Umeå) [https]\n",
      "63: Switzerland [https]\n",
      "64: Taiwan (Taipei) [https]\n",
      "65: Turkey (Denizli) [https]\n",
      "66: Turkey (Istanbul) [https]\n",
      "67: Turkey (Mersin) [https]\n",
      "68: UK (Bristol) [https]\n",
      "69: UK (London 1) [https]\n",
      "70: USA (IA) [https]\n",
      "71: USA (KS) [https]\n",
      "72: USA (MI) [https]\n",
      "73: USA (OH) [https]\n",
      "74: USA (OR) [https]\n",
      "75: USA (TN) [https]\n",
      "76: Uruguay [https]\n",
      "77: (other mirrors)\n",
      "\n",
      "Selection: 1\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "R[write to console]: trying URL 'https://cloud.r-project.org/bin/macosx/el-capitan/contrib/3.6/bnlearn_4.6.1.tgz'\n",
      "\n",
      "R[write to console]: Content type 'application/x-gzip'\n",
      "R[write to console]:  length 2653896 bytes (2.5 MB)\n",
      "\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: =\n",
      "R[write to console]: \n",
      "\n",
      "R[write to console]: downloaded 2.5 MB\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "The downloaded binary packages are in\n",
      "\t/var/folders/40/1cmwtr_5693092wp0mnttzgh0000gn/T//RtmpeSAvU9/downloaded_packages\n"
     ]
    }
   ],
   "source": [
    "listings_fac = rDiscretize(listings_n_cat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>roi</th>\n",
       "      <th>Zestimate</th>\n",
       "      <th>bedrooms</th>\n",
       "      <th>beds</th>\n",
       "      <th>accommodates</th>\n",
       "      <th>amenities_score</th>\n",
       "      <th>publicSchoolScores</th>\n",
       "      <th>closenessToDowntown</th>\n",
       "      <th>price</th>\n",
       "      <th>minimum_nights_avg_ntm</th>\n",
       "      <th>bathrooms</th>\n",
       "      <th>bikeScore</th>\n",
       "      <th>walkScore</th>\n",
       "      <th>transitScore</th>\n",
       "      <th>property_type</th>\n",
       "      <th>neighbourhood</th>\n",
       "      <th>street</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>(2,8]</td>\n",
       "      <td>(5,20]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>(4704.33,19715]</td>\n",
       "      <td>(135,199]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Dorchester</td>\n",
       "      <td>Boston Street</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(-12.0209,9.92724]</td>\n",
       "      <td>(544583,870941]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[2,3]</td>\n",
       "      <td>[12,30]</td>\n",
       "      <td>(3.70142,5.27273]</td>\n",
       "      <td>(4704.33,19715]</td>\n",
       "      <td>(135,199]</td>\n",
       "      <td>(2,28]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Jamaica Plain</td>\n",
       "      <td>Goodrich Road</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[2,3]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>(3.70142,5.27273]</td>\n",
       "      <td>(2400.33,4704.33]</td>\n",
       "      <td>(199,5555]</td>\n",
       "      <td>[1,2]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Charlestown</td>\n",
       "      <td>Salem Street</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[2,3]</td>\n",
       "      <td>[12,30]</td>\n",
       "      <td>(3.47373,3.70142]</td>\n",
       "      <td>(2400.33,4704.33]</td>\n",
       "      <td>[60,135]</td>\n",
       "      <td>(2,28]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Charlestown</td>\n",
       "      <td>Winthrop Street</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[2,3]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>(3.70142,5.27273]</td>\n",
       "      <td>(4704.33,19715]</td>\n",
       "      <td>(135,199]</td>\n",
       "      <td>(2,28]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Allston</td>\n",
       "      <td>Easton Street</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>195</th>\n",
       "      <td>(9.92724,1270.48]</td>\n",
       "      <td>[165887,544583]</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>(2,8]</td>\n",
       "      <td>(5,20]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>[898,2400.33]</td>\n",
       "      <td>(199,5555]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Downtown</td>\n",
       "      <td>Winter Street</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>196</th>\n",
       "      <td>(9.92724,1270.48]</td>\n",
       "      <td>[165887,544583]</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>(3,5]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>[898,2400.33]</td>\n",
       "      <td>(135,199]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Downtown</td>\n",
       "      <td>Bromfield Street</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>197</th>\n",
       "      <td>(9.92724,1270.48]</td>\n",
       "      <td>[165887,544583]</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>(2,8]</td>\n",
       "      <td>(5,20]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>[898,2400.33]</td>\n",
       "      <td>(199,5555]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Downtown</td>\n",
       "      <td>Bromfield Street</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>198</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>(3,5]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>[898,2400.33]</td>\n",
       "      <td>(135,199]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Downtown</td>\n",
       "      <td>Hamilton Place</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>199</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>(2,8]</td>\n",
       "      <td>(5,20]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>[898,2400.33]</td>\n",
       "      <td>(199,5555]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Downtown</td>\n",
       "      <td>Winter Street</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>200 rows × 17 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                     roi             Zestimate bedrooms   beds accommodates  \\\n",
       "0    [-38.1201,-12.0209]  (870941,1.14825e+07]    (2,6]  (2,8]       (5,20]   \n",
       "1     (-12.0209,9.92724]       (544583,870941]    [0,1]  [0,1]        [2,3]   \n",
       "2    [-38.1201,-12.0209]  (870941,1.14825e+07]    [0,1]  [0,1]        [2,3]   \n",
       "3    [-38.1201,-12.0209]  (870941,1.14825e+07]    [0,1]  [0,1]        [2,3]   \n",
       "4    [-38.1201,-12.0209]  (870941,1.14825e+07]    [0,1]  [0,1]        [2,3]   \n",
       "..                   ...                   ...      ...    ...          ...   \n",
       "195    (9.92724,1270.48]       [165887,544583]    (2,6]  (2,8]       (5,20]   \n",
       "196    (9.92724,1270.48]       [165887,544583]    (1,2]  (1,2]        (3,5]   \n",
       "197    (9.92724,1270.48]       [165887,544583]    (2,6]  (2,8]       (5,20]   \n",
       "198  [-38.1201,-12.0209]  (870941,1.14825e+07]    (1,2]  (1,2]        (3,5]   \n",
       "199  [-38.1201,-12.0209]  (870941,1.14825e+07]    (2,6]  (2,8]       (5,20]   \n",
       "\n",
       "    amenities_score publicSchoolScores closenessToDowntown       price  \\\n",
       "0           (30,33]     [2.75,3.47373]     (4704.33,19715]   (135,199]   \n",
       "1           [12,30]  (3.70142,5.27273]     (4704.33,19715]   (135,199]   \n",
       "2           (30,33]  (3.70142,5.27273]   (2400.33,4704.33]  (199,5555]   \n",
       "3           [12,30]  (3.47373,3.70142]   (2400.33,4704.33]    [60,135]   \n",
       "4           (30,33]  (3.70142,5.27273]     (4704.33,19715]   (135,199]   \n",
       "..              ...                ...                 ...         ...   \n",
       "195         (30,33]     [2.75,3.47373]       [898,2400.33]  (199,5555]   \n",
       "196         (30,33]     [2.75,3.47373]       [898,2400.33]   (135,199]   \n",
       "197         (30,33]     [2.75,3.47373]       [898,2400.33]  (199,5555]   \n",
       "198         (30,33]     [2.75,3.47373]       [898,2400.33]   (135,199]   \n",
       "199         (30,33]     [2.75,3.47373]       [898,2400.33]  (199,5555]   \n",
       "\n",
       "    minimum_nights_avg_ntm bathrooms  bikeScore  walkScore  transitScore  \\\n",
       "0                 (28,353]       1.0          3          5             3   \n",
       "1                   (2,28]       1.0          4          5             3   \n",
       "2                    [1,2]       1.0          3          4             4   \n",
       "3                   (2,28]       1.0          3          5             4   \n",
       "4                   (2,28]       1.0          4          4             3   \n",
       "..                     ...       ...        ...        ...           ...   \n",
       "195               (28,353]       2.0          3          5             5   \n",
       "196               (28,353]       1.0          3          5             5   \n",
       "197               (28,353]       2.0          3          5             5   \n",
       "198               (28,353]       1.0          3          5             5   \n",
       "199               (28,353]       2.0          3          5             5   \n",
       "\n",
       "    property_type  neighbourhood            street  \n",
       "0       Apartment     Dorchester     Boston Street  \n",
       "1       Apartment  Jamaica Plain     Goodrich Road  \n",
       "2       Apartment    Charlestown      Salem Street  \n",
       "3       Apartment    Charlestown   Winthrop Street  \n",
       "4       Apartment        Allston     Easton Street  \n",
       "..            ...            ...               ...  \n",
       "195     Apartment       Downtown     Winter Street  \n",
       "196     Apartment       Downtown  Bromfield Street  \n",
       "197     Apartment       Downtown  Bromfield Street  \n",
       "198     Apartment       Downtown    Hamilton Place  \n",
       "199     Apartment       Downtown     Winter Street  \n",
       "\n",
       "[200 rows x 17 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_discrete = listings_fac.reset_index(drop=True).join(listing_cat)\n",
    "data_discrete"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>roi</th>\n",
       "      <th>Zestimate</th>\n",
       "      <th>bedrooms</th>\n",
       "      <th>beds</th>\n",
       "      <th>accommodates</th>\n",
       "      <th>amenities_score</th>\n",
       "      <th>publicSchoolScores</th>\n",
       "      <th>closenessToDowntown</th>\n",
       "      <th>price</th>\n",
       "      <th>minimum_nights_avg_ntm</th>\n",
       "      <th>bathrooms</th>\n",
       "      <th>bikeScore</th>\n",
       "      <th>walkScore</th>\n",
       "      <th>transitScore</th>\n",
       "      <th>property_type</th>\n",
       "      <th>neighbourhood</th>\n",
       "      <th>street</th>\n",
       "      <th>mobility</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>(2,8]</td>\n",
       "      <td>(5,20]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>(4704.33,19715]</td>\n",
       "      <td>(135,199]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Dorchester</td>\n",
       "      <td>Boston Street</td>\n",
       "      <td>3.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(-12.0209,9.92724]</td>\n",
       "      <td>(544583,870941]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[2,3]</td>\n",
       "      <td>[12,30]</td>\n",
       "      <td>(3.70142,5.27273]</td>\n",
       "      <td>(4704.33,19715]</td>\n",
       "      <td>(135,199]</td>\n",
       "      <td>(2,28]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Jamaica Plain</td>\n",
       "      <td>Goodrich Road</td>\n",
       "      <td>4.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[2,3]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>(3.70142,5.27273]</td>\n",
       "      <td>(2400.33,4704.33]</td>\n",
       "      <td>(199,5555]</td>\n",
       "      <td>[1,2]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Charlestown</td>\n",
       "      <td>Salem Street</td>\n",
       "      <td>3.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[2,3]</td>\n",
       "      <td>[12,30]</td>\n",
       "      <td>(3.47373,3.70142]</td>\n",
       "      <td>(2400.33,4704.33]</td>\n",
       "      <td>[60,135]</td>\n",
       "      <td>(2,28]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Charlestown</td>\n",
       "      <td>Winthrop Street</td>\n",
       "      <td>4.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>[2,3]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>(3.70142,5.27273]</td>\n",
       "      <td>(4704.33,19715]</td>\n",
       "      <td>(135,199]</td>\n",
       "      <td>(2,28]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Allston</td>\n",
       "      <td>Easton Street</td>\n",
       "      <td>3.666667</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>195</th>\n",
       "      <td>(9.92724,1270.48]</td>\n",
       "      <td>[165887,544583]</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>(2,8]</td>\n",
       "      <td>(5,20]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>[898,2400.33]</td>\n",
       "      <td>(199,5555]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Downtown</td>\n",
       "      <td>Winter Street</td>\n",
       "      <td>4.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>196</th>\n",
       "      <td>(9.92724,1270.48]</td>\n",
       "      <td>[165887,544583]</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>(3,5]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>[898,2400.33]</td>\n",
       "      <td>(135,199]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Downtown</td>\n",
       "      <td>Bromfield Street</td>\n",
       "      <td>4.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>197</th>\n",
       "      <td>(9.92724,1270.48]</td>\n",
       "      <td>[165887,544583]</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>(2,8]</td>\n",
       "      <td>(5,20]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>[898,2400.33]</td>\n",
       "      <td>(199,5555]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Downtown</td>\n",
       "      <td>Bromfield Street</td>\n",
       "      <td>4.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>198</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>(3,5]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>[898,2400.33]</td>\n",
       "      <td>(135,199]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Downtown</td>\n",
       "      <td>Hamilton Place</td>\n",
       "      <td>4.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>199</th>\n",
       "      <td>[-38.1201,-12.0209]</td>\n",
       "      <td>(870941,1.14825e+07]</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>(2,8]</td>\n",
       "      <td>(5,20]</td>\n",
       "      <td>(30,33]</td>\n",
       "      <td>[2.75,3.47373]</td>\n",
       "      <td>[898,2400.33]</td>\n",
       "      <td>(199,5555]</td>\n",
       "      <td>(28,353]</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>Downtown</td>\n",
       "      <td>Winter Street</td>\n",
       "      <td>4.333333</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>200 rows × 18 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                     roi             Zestimate bedrooms   beds accommodates  \\\n",
       "0    [-38.1201,-12.0209]  (870941,1.14825e+07]    (2,6]  (2,8]       (5,20]   \n",
       "1     (-12.0209,9.92724]       (544583,870941]    [0,1]  [0,1]        [2,3]   \n",
       "2    [-38.1201,-12.0209]  (870941,1.14825e+07]    [0,1]  [0,1]        [2,3]   \n",
       "3    [-38.1201,-12.0209]  (870941,1.14825e+07]    [0,1]  [0,1]        [2,3]   \n",
       "4    [-38.1201,-12.0209]  (870941,1.14825e+07]    [0,1]  [0,1]        [2,3]   \n",
       "..                   ...                   ...      ...    ...          ...   \n",
       "195    (9.92724,1270.48]       [165887,544583]    (2,6]  (2,8]       (5,20]   \n",
       "196    (9.92724,1270.48]       [165887,544583]    (1,2]  (1,2]        (3,5]   \n",
       "197    (9.92724,1270.48]       [165887,544583]    (2,6]  (2,8]       (5,20]   \n",
       "198  [-38.1201,-12.0209]  (870941,1.14825e+07]    (1,2]  (1,2]        (3,5]   \n",
       "199  [-38.1201,-12.0209]  (870941,1.14825e+07]    (2,6]  (2,8]       (5,20]   \n",
       "\n",
       "    amenities_score publicSchoolScores closenessToDowntown       price  \\\n",
       "0           (30,33]     [2.75,3.47373]     (4704.33,19715]   (135,199]   \n",
       "1           [12,30]  (3.70142,5.27273]     (4704.33,19715]   (135,199]   \n",
       "2           (30,33]  (3.70142,5.27273]   (2400.33,4704.33]  (199,5555]   \n",
       "3           [12,30]  (3.47373,3.70142]   (2400.33,4704.33]    [60,135]   \n",
       "4           (30,33]  (3.70142,5.27273]     (4704.33,19715]   (135,199]   \n",
       "..              ...                ...                 ...         ...   \n",
       "195         (30,33]     [2.75,3.47373]       [898,2400.33]  (199,5555]   \n",
       "196         (30,33]     [2.75,3.47373]       [898,2400.33]   (135,199]   \n",
       "197         (30,33]     [2.75,3.47373]       [898,2400.33]  (199,5555]   \n",
       "198         (30,33]     [2.75,3.47373]       [898,2400.33]   (135,199]   \n",
       "199         (30,33]     [2.75,3.47373]       [898,2400.33]  (199,5555]   \n",
       "\n",
       "    minimum_nights_avg_ntm bathrooms  bikeScore  walkScore  transitScore  \\\n",
       "0                 (28,353]       1.0          3          5             3   \n",
       "1                   (2,28]       1.0          4          5             3   \n",
       "2                    [1,2]       1.0          3          4             4   \n",
       "3                   (2,28]       1.0          3          5             4   \n",
       "4                   (2,28]       1.0          4          4             3   \n",
       "..                     ...       ...        ...        ...           ...   \n",
       "195               (28,353]       2.0          3          5             5   \n",
       "196               (28,353]       1.0          3          5             5   \n",
       "197               (28,353]       2.0          3          5             5   \n",
       "198               (28,353]       1.0          3          5             5   \n",
       "199               (28,353]       2.0          3          5             5   \n",
       "\n",
       "    property_type  neighbourhood            street  mobility  \n",
       "0       Apartment     Dorchester     Boston Street  3.666667  \n",
       "1       Apartment  Jamaica Plain     Goodrich Road  4.000000  \n",
       "2       Apartment    Charlestown      Salem Street  3.666667  \n",
       "3       Apartment    Charlestown   Winthrop Street  4.000000  \n",
       "4       Apartment        Allston     Easton Street  3.666667  \n",
       "..            ...            ...               ...       ...  \n",
       "195     Apartment       Downtown     Winter Street  4.333333  \n",
       "196     Apartment       Downtown  Bromfield Street  4.333333  \n",
       "197     Apartment       Downtown  Bromfield Street  4.333333  \n",
       "198     Apartment       Downtown    Hamilton Place  4.333333  \n",
       "199     Apartment       Downtown     Winter Street  4.333333  \n",
       "\n",
       "[200 rows x 18 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_discrete['mobility']= data_discrete[['bikeScore', 'walkScore', 'transitScore']].mean(axis=1)\n",
    "data_discrete"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "nodes = ['amenities_score','bathrooms','bedrooms','Zestimate','neighbourhood',\n",
    "         'roi','price','minimum_nights_avg_ntm','publicSchoolScores', 'mobility',\n",
    "         'property_type','closenessToDowntown','accommodates','beds','street']\n",
    "edges = [('amenities_score','price'),\n",
    "         ('bathrooms','price'),\n",
    "         ('bathrooms','Zestimate'),\n",
    "         ('bedrooms','price'),\n",
    "         ('bedrooms','beds'),\n",
    "         ('bedrooms','accommodates'),\n",
    "         ('bedrooms','Zestimate'),\n",
    "         ('property_type','price'),\n",
    "         ('property_type','Zestimate'),\n",
    "         ('price','roi'),\n",
    "         ('mobility','Zestimate'),\n",
    "         ('Zestimate','price'),\n",
    "         ('mobility','price'),\n",
    "         ('Zestimate','roi'),\n",
    "         ('beds','price'),\n",
    "         ('accommodates','price'),\n",
    "         ('neighbourhood','street'),\n",
    "         ('street','price'),\n",
    "         ('street','mobility'),\n",
    "         ('street','publicSchoolScores'),\n",
    "         ('publicSchoolScores','Zestimate'),\n",
    "         ('street','Zestimate'),\n",
    "         ('street','closenessToDowntown'),\n",
    "         ('closenessToDowntown','price'),                    \n",
    "         ('minimum_nights_avg_ntm','price')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pgmpy.models import BayesianModel\n",
    "graph = BayesianModel()\n",
    "\n",
    "graph.add_nodes_from(nodes)\n",
    "graph.add_edges_from(edges)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 14400x7200 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import networkx as nx\n",
    "import pydot\n",
    "\n",
    "DAG_11 = nx.DiGraph()\n",
    "DAG_11.add_edges_from(edges)\n",
    "pos_11 = nx.nx_pydot.graphviz_layout(DAG_11, prog='dot')\n",
    "nx.draw(DAG_11, pos=pos_11, with_labels=True)\n",
    "plt.figure(figsize=(200,100))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def adjacencyMatrix(edges):\n",
    "    df = pd.DataFrame(edges, columns =['source','target'], dtype = float) \n",
    "    df = pd.crosstab(df.source, df.target)\n",
    "    idx = df.columns.union(df.index)\n",
    "    df2 = df.reindex(index = idx, columns=idx, fill_value=0)\n",
    "    df2.index.name = \"source\"\n",
    "    df2 = df2.rename_axis('target')\n",
    "    return df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "adjMat = adjacencyMatrix(edges)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# adjMat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_discrete = data_discrete.astype(object)\n",
    "model ={'adjmat':adjMat, 'model':graph}\n",
    "\n",
    "model_update = bnlearn.parameter_learning.fit(model, data_discrete, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('mymodel.pickle', 'wb') as f:\n",
    "    pickle.dump(model_update, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "now = 2020-12-17 00:11:58.231171\n",
      "date and time = 17/12/2020 00:11:58\n"
     ]
    }
   ],
   "source": [
    "from datetime import datetime\n",
    "\n",
    "# datetime object containing current date and time\n",
    "now = datetime.now()\n",
    " \n",
    "print(\"now =\", now)\n",
    "\n",
    "# dd/mm/YY H:M:S\n",
    "dt_string = now.strftime(\"%d/%m/%Y %H:%M:%S\")\n",
    "print(\"date and time =\", dt_string)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "amenities_score  : 3\n",
      "bathrooms  : 5\n",
      "bedrooms  : 3\n",
      "Zestimate  : 3\n",
      "neighbourhood  : 20\n",
      "roi  : 3\n",
      "price  : 3\n",
      "minimum_nights_avg_ntm  : 3\n",
      "publicSchoolScores  : 3\n",
      "mobility  : 9\n",
      "property_type  : 2\n",
      "closenessToDowntown  : 3\n",
      "accommodates  : 3\n",
      "beds  : 3\n",
      "street  : 147\n"
     ]
    }
   ],
   "source": [
    "labels ={}\n",
    "for i in list(model_update['model'].nodes):\n",
    "    labels[i] = model_update['model'].get_cpds(i).state_names[i]\n",
    "    \n",
    "    print(i,\" :\",len(labels[i]))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "params_amenities_score = torch.tensor(model_update['model'].get_cpds('amenities_score').values.T)\n",
    "params_bathrooms = torch.tensor(model_update['model'].get_cpds('bathrooms').values.T)\n",
    "params_bedrooms = torch.tensor(model_update['model'].get_cpds('bedrooms').values.T)\n",
    "params_minimum_nights_avg_ntm = torch.tensor(model_update['model'].get_cpds('minimum_nights_avg_ntm').values.T)\n",
    "params_property_type = torch.tensor(model_update['model'].get_cpds('property_type').values.T)\n",
    "params_neighbourhood = torch.tensor(model_update['model'].get_cpds('neighbourhood').values.T)\n",
    "params_accommodates = torch.tensor(model_update['model'].get_cpds('accommodates').values.T)\n",
    "params_beds = torch.tensor(model_update['model'].get_cpds('beds').values.T)\n",
    "params_closenessToDowntown = torch.tensor(model_update['model'].get_cpds('closenessToDowntown').values.T)\n",
    "params_publicSchoolScores = torch.tensor(model_update['model'].get_cpds('publicSchoolScores').values.T)\n",
    "params_Zestimate = torch.tensor(model_update['model'].get_cpds('Zestimate').values.T)\n",
    "params_price = torch.tensor(model_update['model'].get_cpds('price').values.T)\n",
    "params_roi = torch.tensor(model_update['model'].get_cpds('roi').values.T)\n",
    "params_mobility = torch.tensor(model_update['model'].get_cpds('mobility').values.T)\n",
    "params_street = torch.tensor(model_update['model'].get_cpds('street').values.T)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "params_amenities_score \t\t\t torch.Size([3])\n",
      "params_bathrooms \t\t\t torch.Size([5])\n",
      "params_bedrooms \t\t\t torch.Size([3])\n",
      "params_minimum_nights_avg_ntm \t\t torch.Size([3])\n",
      "params_property_type \t\t\t torch.Size([2])\n",
      "params_neighbourhood \t\t\t torch.Size([20])\n",
      "params_accommodates \t\t\t torch.Size([3, 3])\n",
      "params_beds \t\t\t\t torch.Size([3, 3])\n",
      "params_closenessToDowntown \t\t torch.Size([147, 3])\n",
      "params_publicSchoolScores \t\t torch.Size([147, 3])\n",
      "params_mobility \t\t\t torch.Size([147, 9])\n",
      "params_street \t\t\t\t torch.Size([20, 147])\n",
      "params_roi \t\t\t\t torch.Size([3, 3, 3])\n",
      "params_Zestimate \t\t\t torch.Size([147, 3, 2, 9, 3, 5, 3])\n",
      "params_price \t\t\t\t torch.Size([147, 2, 9, 3, 3, 3, 3, 5, 3, 3, 3, 3])\n"
     ]
    }
   ],
   "source": [
    "print(\"params_amenities_score \\t\\t\\t\",params_amenities_score.shape)\n",
    "print(\"params_bathrooms \\t\\t\\t\",params_bathrooms.shape)\n",
    "print(\"params_bedrooms \\t\\t\\t\",params_bedrooms.shape)\n",
    "print(\"params_minimum_nights_avg_ntm \\t\\t\",params_minimum_nights_avg_ntm.shape)\n",
    "print(\"params_property_type \\t\\t\\t\",params_property_type.shape)\n",
    "print(\"params_neighbourhood \\t\\t\\t\",params_neighbourhood.shape)\n",
    "\n",
    "print(\"params_accommodates \\t\\t\\t\",params_accommodates.shape)\n",
    "print(\"params_beds \\t\\t\\t\\t\",params_beds.shape)\n",
    "\n",
    "print(\"params_closenessToDowntown \\t\\t\",params_closenessToDowntown.shape)\n",
    "print(\"params_publicSchoolScores \\t\\t\",params_publicSchoolScores.shape)\n",
    "print(\"params_mobility \\t\\t\\t\",params_mobility.shape)\n",
    "print(\"params_street \\t\\t\\t\\t\",params_street.shape)\n",
    "print(\"params_roi \\t\\t\\t\\t\",params_roi.shape)\n",
    "print(\"params_Zestimate \\t\\t\\t\",params_Zestimate.shape)\n",
    "print(\"params_price \\t\\t\\t\\t\",params_price.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pyroModel():\n",
    "    amenities_score = pyro.sample('amenities_score', dist.Categorical(params_amenities_score))\n",
    "    bathrooms = pyro.sample('bathrooms', dist.Categorical(params_bathrooms))\n",
    "    bedrooms = pyro.sample('bedrooms', dist.Categorical(params_bedrooms))\n",
    "    minimum_nights_avg_ntm = pyro.sample('minimum_nights_avg_ntm', dist.Categorical(params_minimum_nights_avg_ntm))\n",
    "    property_type = pyro.sample('property_type', dist.Categorical(params_property_type))\n",
    "    neighbourhood = pyro.sample('neighbourhood', dist.Categorical(params_neighbourhood))\n",
    "    \n",
    "    accommodates = pyro.sample('accommodates', dist.Categorical(params_accommodates[bedrooms]))\n",
    "    beds = pyro.sample('beds', dist.Categorical(params_beds[bedrooms]))\n",
    "    street = pyro.sample('street', dist.Categorical(params_street[neighbourhood]))\n",
    "    closenessToDowntown = pyro.sample('closenessToDowntown', dist.Categorical(params_closenessToDowntown[street]))\n",
    "    publicSchoolScores = pyro.sample('publicSchoolScores', dist.Categorical(params_publicSchoolScores[street]))\n",
    "    mobility = pyro.sample('mobility', dist.Categorical(params_mobility[street]))\n",
    "    \n",
    "    Zestimate = pyro.sample('Zestimate',dist.Categorical(params_Zestimate[street][publicSchoolScores]\n",
    "                                                         [property_type][mobility][bedrooms][bathrooms]))\n",
    "    \n",
    "    price = pyro.sample('price', dist.Categorical(params_price[street][property_type][mobility]\n",
    "                                                  [minimum_nights_avg_ntm][closenessToDowntown][beds][bedrooms]\n",
    "                                                  [bathrooms][amenities_score][accommodates][Zestimate]))\n",
    "    roi = pyro.sample('roi', dist.Categorical(params_roi[price][Zestimate]))\n",
    "    \n",
    "    return(amenities_score,bathrooms,bedrooms,minimum_nights_avg_ntm,property_type,neighbourhood,\n",
    "           accommodates,beds,closenessToDowntown,publicSchoolScores,street,mobility,\n",
    "           Zestimate,price,roi)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "def UsingDo(data,value):\n",
    "    intervention_model = pyro.do(pyroModel, data={data: torch.tensor(value)})\n",
    "    posterior = pyro.infer.Importance(intervention_model, num_samples=5000).run()\n",
    "    marginal = EmpiricalMarginal(posterior,\"roi\")\n",
    "    samples = [marginal().item() for _ in range(1000)]\n",
    "    \n",
    "    unique, counts = np.unique(samples, return_counts=True)\n",
    "    prob = counts[1]/sum(counts)\n",
    "  \n",
    "    return(prob)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "def UsingConditional(data,value):\n",
    "    conditioned_model = pyro.condition(pyroModel, data={data: torch.tensor(value)})\n",
    "    posterior = pyro.infer.Importance(conditioned_model, num_samples=5000).run()\n",
    "    marginal = EmpiricalMarginal(posterior,\"roi\")\n",
    "    samples = [marginal().item() for _ in range(1000)]\n",
    "    \n",
    "    unique, counts = np.unique(samples, return_counts=True)\n",
    "    prob = counts[1]/sum(counts)\n",
    "  \n",
    "    return(prob)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1,2] : 0.323\n",
      "(2,6] : 0.329\n",
      "[0,1] : 0.327\n",
      "\n",
      "(1,2] : 0.348\n",
      "(2,6] : 0.379\n",
      "[0,1] : 0.329\n"
     ]
    }
   ],
   "source": [
    "pyro.set_rng_seed(10)\n",
    "#Probability of ROI fixing the bedrooms using do operator\n",
    "for i in range(len(labels['bedrooms'])):\n",
    "    val = UsingDo('bedrooms', i)\n",
    "    print(labels['bedrooms'][i],':',val)\n",
    "print()\n",
    "pyro.set_rng_seed(10)\n",
    "for i in range(len(labels['bedrooms'])):\n",
    "    val = UsingConditional('bedrooms', i)\n",
    "    print(labels['bedrooms'][i],':',val)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "3 - 5 rooms give best roi"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.0 : 0.323\n",
      "1.5 : 0.329\n",
      "2.0 : 0.326\n",
      "2.5 : 0.325\n",
      "3.0 : 0.341\n",
      "\n",
      "1.0 : 0.347\n",
      "1.5 : 0.379\n",
      "2.0 : 0.331\n",
      "2.5 : 0.327\n",
      "3.0 : 0.298\n"
     ]
    }
   ],
   "source": [
    "pyro.set_rng_seed(10)\n",
    "#Probability of ROI fixing the bedrooms using do operator\n",
    "for i in range(len(labels['bathrooms'])):\n",
    "    val = UsingDo('bathrooms', i)\n",
    "    print(labels['bathrooms'][i],':',val)\n",
    "print()\n",
    "pyro.set_rng_seed(10)\n",
    "for i in range(len(labels['bathrooms'])):\n",
    "    val = UsingConditional('bathrooms', i)\n",
    "    print(labels['bathrooms'][i],':',val)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(30,33] : 0.323\n",
      "(33,52] : 0.329\n",
      "[12,30] : 0.326\n",
      "\n",
      "(30,33] : 0.348\n",
      "(33,52] : 0.377\n",
      "[12,30] : 0.331\n"
     ]
    }
   ],
   "source": [
    "pyro.set_rng_seed(10)\n",
    "#Probability of ROI fixing the bedrooms using do operator\n",
    "for i in range(len(labels['amenities_score'])):\n",
    "    val = UsingDo('amenities_score', i)\n",
    "    print(labels['amenities_score'][i],':',val)\n",
    "print()\n",
    "pyro.set_rng_seed(10)\n",
    "for i in range(len(labels['amenities_score'])):\n",
    "    val = UsingConditional('amenities_score', i)\n",
    "    print(labels['amenities_score'][i],':',val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Allston : 0.323\n",
      "Back Bay : 0.33\n",
      "Bay Village : 0.326\n",
      "Beacon Hill : 0.326\n",
      "Brighton : 0.34\n",
      "Charlestown : 0.318\n",
      "Chinatown : 0.339\n",
      "Dorchester : 0.324\n",
      "Downtown : 0.339\n",
      "East Boston : 0.344\n",
      "Fenway : 0.323\n",
      "Hyde Park : 0.303\n",
      "Jamaica Plain : 0.332\n",
      "Mission Hill : 0.33\n",
      "North End : 0.326\n",
      "Roslindale : 0.335\n",
      "Roxbury : 0.357\n",
      "South Boston : 0.318\n",
      "South End : 0.378\n",
      "West Roxbury : 0.342\n",
      "\n",
      "Allston : 0.349\n",
      "Back Bay : 0.38\n",
      "Bay Village : 0.33\n",
      "Beacon Hill : 0.327\n",
      "Brighton : 0.297\n",
      "Charlestown : 0.321\n",
      "Chinatown : 0.337\n",
      "Dorchester : 0.337\n",
      "Downtown : 0.324\n",
      "East Boston : 0.354\n",
      "Fenway : 0.332\n",
      "Hyde Park : 0.351\n",
      "Jamaica Plain : 0.34\n",
      "Mission Hill : 0.32\n",
      "North End : 0.307\n",
      "Roslindale : 0.324\n",
      "Roxbury : 0.37\n",
      "South Boston : 0.368\n",
      "South End : 0.322\n",
      "West Roxbury : 0.325\n"
     ]
    }
   ],
   "source": [
    "pyro.set_rng_seed(10)\n",
    "#Probability of ROI fixing the bedrooms using do operator\n",
    "for i in range(len(labels['neighbourhood'])):\n",
    "    val = UsingDo('neighbourhood', i)\n",
    "    print(labels['neighbourhood'][i],':',val)\n",
    "print()\n",
    "pyro.set_rng_seed(10)\n",
    "for i in range(len(labels['neighbourhood'])):\n",
    "    val = UsingConditional('neighbourhood', i)\n",
    "    print(labels['neighbourhood'][i],':',val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2400.33,4704.33] : 0.323\n",
      "(4704.33,19715] : 0.329\n",
      "[898,2400.33] : 0.326\n",
      "\n",
      "(2400.33,4704.33] : 0.371\n",
      "(4704.33,19715] : 0.335\n",
      "[898,2400.33] : 0.33\n"
     ]
    }
   ],
   "source": [
    "pyro.set_rng_seed(10)\n",
    "#Probability of ROI fixing the bedrooms using do operator\n",
    "for i in range(len(labels['closenessToDowntown'])):\n",
    "    val = UsingDo('closenessToDowntown', i)\n",
    "    print(labels['closenessToDowntown'][i],':',val)\n",
    "print()\n",
    "pyro.set_rng_seed(10)\n",
    "for i in range(len(labels['closenessToDowntown'])):\n",
    "    val = UsingConditional('closenessToDowntown', i)\n",
    "    print(labels['closenessToDowntown'][i],':',val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(544583,870941] : 0.34\n",
      "(870941,1.14825e+07] : 0.281\n",
      "[165887,544583] : 0.383\n",
      "\n",
      "(544583,870941] : 0.348\n",
      "(870941,1.14825e+07] : 0.291\n",
      "[165887,544583] : 0.384\n"
     ]
    }
   ],
   "source": [
    "pyro.set_rng_seed(10)\n",
    "#Probability of ROI fixing the bedrooms using do operator\n",
    "for i in range(len(labels['Zestimate'])):\n",
    "    val = UsingDo('Zestimate', i)\n",
    "    print(labels['Zestimate'][i],':',val)\n",
    "print()\n",
    "pyro.set_rng_seed(10)\n",
    "for i in range(len(labels['Zestimate'])):\n",
    "    val = UsingConditional('Zestimate', i)\n",
    "    print(labels['Zestimate'][i],':',val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.0 : 0.323\n",
      "2.3333333333333335 : 0.329\n",
      "2.6666666666666665 : 0.326\n",
      "3.0 : 0.325\n",
      "3.3333333333333335 : 0.341\n",
      "3.6666666666666665 : 0.317\n",
      "4.0 : 0.34\n",
      "4.333333333333333 : 0.324\n",
      "4.666666666666667 : 0.337\n",
      "\n",
      "2.0 : 0.346\n",
      "2.3333333333333335 : 0.337\n",
      "2.6666666666666665 : 0.326\n",
      "3.0 : 0.364\n",
      "3.3333333333333335 : 0.34\n",
      "3.6666666666666665 : 0.325\n",
      "4.0 : 0.339\n",
      "4.333333333333333 : 0.333\n",
      "4.666666666666667 : 0.314\n"
     ]
    }
   ],
   "source": [
    "pyro.set_rng_seed(10)\n",
    "#Probability of ROI fixing the bedrooms using do operator\n",
    "for i in range(len(labels['mobility'])):\n",
    "    val = UsingDo('mobility', i)\n",
    "    print(labels['mobility'][i],':',val)\n",
    "print()\n",
    "pyro.set_rng_seed(10)\n",
    "for i in range(len(labels['mobility'])):\n",
    "    val = UsingConditional('mobility', i)\n",
    "    print(labels['mobility'][i],':',val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Alameda Road : 0.323\n",
      "Appleton Street : 0.328\n",
      "Argyle Street : 0.326\n",
      "Armington Street : 0.325\n",
      "Ashburton Place : 0.339\n",
      "Beacon Street : 0.318\n",
      "Beech Street : 0.34\n",
      "Bellingham Place : 0.324\n",
      "Bennington Street : 0.337\n",
      "Blue Hill Avenue : 0.345\n",
      "Bolton Street : 0.324\n",
      "Bond Street : 0.304\n",
      "Boston Street : 0.331\n",
      "Bostonia Avenue : 0.331\n",
      "Bremen Street : 0.325\n",
      "Brent Street : 0.334\n",
      "Brighton Avenue : 0.357\n",
      "Brighton Street : 0.319\n",
      "Brimmer Street : 0.377\n",
      "Bromfield Street : 0.343\n",
      "Brookline Avenue : 0.312\n",
      "Brookside Avenue : 0.33\n",
      "Bunker Hill Street : 0.358\n",
      "Business Street : 0.328\n",
      "Buttonwood Street : 0.342\n",
      "Byron Street : 0.321\n",
      "Cedar Lane Way : 0.342\n",
      "Centre Street : 0.332\n",
      "Charles Street : 0.337\n",
      "Charles Street South : 0.319\n",
      "Chelsea Street : 0.361\n",
      "Clive Street : 0.33\n",
      "Colburn Street : 0.335\n",
      "Columbia Road : 0.342\n",
      "Columbus Avenue : 0.339\n",
      "Commonwealth Avenue : 0.342\n",
      "Commonwealth Avenue Frontage Road Wb : 0.345\n",
      "Concord Square : 0.32\n",
      "Corey Road : 0.344\n",
      "Cowing Street : 0.351\n",
      "Crawford Street : 0.332\n",
      "Cross Street : 0.332\n",
      "D Street : 0.334\n",
      "Delle Avenue : 0.332\n",
      "Derne Street : 0.356\n",
      "Dorchester Avenue : 0.319\n",
      "Dorchester Street : 0.342\n",
      "Durham Street : 0.345\n",
      "Dwight Street : 0.345\n",
      "East Fourth Street : 0.324\n",
      "East Springfield Street : 0.334\n",
      "Easton Street : 0.34\n",
      "Edgerly Road : 0.335\n",
      "Ellis Street : 0.305\n",
      "G Street : 0.33\n",
      "Garden Street : 0.305\n",
      "Geneva Avenue : 0.318\n",
      "Glenley Terrace : 0.355\n",
      "Gold Street : 0.336\n",
      "Goodrich Road : 0.334\n",
      "Green Street : 0.345\n",
      "Greenough Lane : 0.354\n",
      "Greenville Street : 0.319\n",
      "Greenwich Park : 0.333\n",
      "Hamilton Place : 0.325\n",
      "Hamilton Street : 0.325\n",
      "Hanover Street : 0.308\n",
      "Hanson Street : 0.337\n",
      "Hardwick Street : 0.339\n",
      "Harrison Avenue : 0.313\n",
      "Haynes Street : 0.312\n",
      "Hooten Court : 0.355\n",
      "Huntington Avenue : 0.32\n",
      "I Street : 0.341\n",
      "Isabella Street : 0.339\n",
      "K Street : 0.372\n",
      "Kilsyth Road : 0.347\n",
      "Kinross Road : 0.368\n",
      "Knapp Street : 0.366\n",
      "Larch Street : 0.331\n",
      "Lawrence Street : 0.318\n",
      "Lexington Street : 0.317\n",
      "Leyden Street : 0.34\n",
      "M Street : 0.328\n",
      "Marcella Street : 0.325\n",
      "Marginal Street : 0.324\n",
      "Marlborough Street : 0.339\n",
      "Massachusetts Avenue : 0.372\n",
      "Mendell Way : 0.326\n",
      "Meridian Street : 0.323\n",
      "Monadnock Street : 0.318\n",
      "Montebello Road : 0.33\n",
      "Mount Vernon Street : 0.332\n",
      "Mozart Street : 0.318\n",
      "Neponset Avenue : 0.325\n",
      "Newbury Street : 0.336\n",
      "North Street : 0.331\n",
      "Norway Street : 0.346\n",
      "Orchardhill Road : 0.324\n",
      "Park Lane : 0.341\n",
      "Parkman Street : 0.334\n",
      "Peterborough Street : 0.326\n",
      "Phillips Street : 0.327\n",
      "Pinckney Street : 0.307\n",
      "Princeton Street : 0.309\n",
      "Quint Avenue : 0.359\n",
      "Revere Street : 0.344\n",
      "Rutland Street : 0.32\n",
      "Sagamore Street : 0.361\n",
      "Saint Botolph Street : 0.344\n",
      "Saint Germain Street : 0.325\n",
      "Salem Street : 0.337\n",
      "Saratoga Street : 0.333\n",
      "Sawyer Avenue : 0.348\n",
      "School Street : 0.361\n",
      "Sentry Hill Place : 0.338\n",
      "Silver Street : 0.333\n",
      "Smith Court : 0.363\n",
      "Soley Street : 0.301\n",
      "South Huntington Avenue : 0.335\n",
      "Strathmore Road : 0.333\n",
      "Sullivan Street : 0.322\n",
      "Sumner Street : 0.38\n",
      "Telegraph Street : 0.323\n",
      "Temple Street : 0.321\n",
      "Thornton Street : 0.298\n",
      "Torpie Street : 0.334\n",
      "Tremont Street : 0.304\n",
      "Trenton Street : 0.348\n",
      "Tudor Street : 0.346\n",
      "Tyler Street : 0.35\n",
      "Union Park : 0.349\n",
      "Upton Street : 0.344\n",
      "Waltham Street : 0.35\n",
      "Washington Street : 0.333\n",
      "Waverly Street : 0.314\n",
      "Webster Street : 0.34\n",
      "Weld Hill Street : 0.345\n",
      "West Canton Street : 0.345\n",
      "West Cedar Street : 0.341\n",
      "West Second Street : 0.325\n",
      "West Springfield Street : 0.33\n",
      "West Third Street : 0.33\n",
      "Winter Street : 0.342\n",
      "Winthrop Street : 0.319\n",
      "Worcester Square : 0.349\n",
      "Worcester Street : 0.362\n",
      "0.38\n",
      "\n",
      "Alameda Road : 0.36\n",
      "Appleton Street : 0.349\n",
      "Argyle Street : 0.313\n",
      "Armington Street : 0.332\n",
      "Ashburton Place : 0.297\n",
      "Beacon Street : 0.321\n",
      "Beech Street : 0.34\n",
      "Bellingham Place : 0.325\n",
      "Bennington Street : 0.357\n",
      "Blue Hill Avenue : 0.336\n",
      "Bolton Street : 0.346\n",
      "Bond Street : 0.33\n",
      "Boston Street : 0.326\n",
      "Bostonia Avenue : 0.302\n",
      "Bremen Street : 0.321\n",
      "Brent Street : 0.345\n",
      "Brighton Avenue : 0.335\n",
      "Brighton Street : 0.319\n",
      "Brimmer Street : 0.357\n",
      "Bromfield Street : 0.353\n",
      "Brookline Avenue : 0.34\n",
      "Brookside Avenue : 0.329\n",
      "Bunker Hill Street : 0.343\n",
      "Business Street : 0.339\n",
      "Buttonwood Street : 0.355\n",
      "Byron Street : 0.351\n",
      "Cedar Lane Way : 0.334\n",
      "Centre Street : 0.345\n",
      "Charles Street : 0.346\n",
      "Charles Street South : 0.324\n",
      "Chelsea Street : 0.378\n",
      "Clive Street : 0.327\n",
      "Colburn Street : 0.333\n",
      "Columbia Road : 0.353\n",
      "Columbus Avenue : 0.314\n",
      "Commonwealth Avenue : 0.338\n",
      "Commonwealth Avenue Frontage Road Wb : 0.347\n",
      "Concord Square : 0.323\n",
      "Corey Road : 0.354\n",
      "Cowing Street : 0.351\n",
      "Crawford Street : 0.345\n",
      "Cross Street : 0.344\n",
      "D Street : 0.3\n",
      "Delle Avenue : 0.379\n",
      "Derne Street : 0.346\n",
      "Dorchester Avenue : 0.352\n",
      "Dorchester Street : 0.316\n",
      "Durham Street : 0.323\n",
      "Dwight Street : 0.342\n",
      "East Fourth Street : 0.334\n",
      "East Springfield Street : 0.331\n",
      "Easton Street : 0.339\n",
      "Edgerly Road : 0.333\n",
      "Ellis Street : 0.347\n",
      "G Street : 0.332\n",
      "Garden Street : 0.338\n",
      "Geneva Avenue : 0.352\n",
      "Glenley Terrace : 0.362\n",
      "Gold Street : 0.345\n",
      "Goodrich Road : 0.316\n",
      "Green Street : 0.341\n",
      "Greenough Lane : 0.34\n",
      "Greenville Street : 0.309\n",
      "Greenwich Park : 0.371\n",
      "Hamilton Place : 0.339\n",
      "Hamilton Street : 0.333\n",
      "Hanover Street : 0.337\n",
      "Hanson Street : 0.319\n",
      "Hardwick Street : 0.336\n",
      "Harrison Avenue : 0.329\n",
      "Haynes Street : 0.346\n",
      "Hooten Court : 0.31\n",
      "Huntington Avenue : 0.353\n",
      "I Street : 0.338\n",
      "Isabella Street : 0.325\n",
      "K Street : 0.355\n",
      "Kilsyth Road : 0.355\n",
      "Kinross Road : 0.329\n",
      "Knapp Street : 0.351\n",
      "Larch Street : 0.335\n",
      "Lawrence Street : 0.338\n",
      "Lexington Street : 0.323\n",
      "Leyden Street : 0.339\n",
      "M Street : 0.362\n",
      "Marcella Street : 0.295\n",
      "Marginal Street : 0.344\n",
      "Marlborough Street : 0.34\n",
      "Massachusetts Avenue : 0.318\n",
      "Mendell Way : 0.368\n",
      "Meridian Street : 0.314\n",
      "Monadnock Street : 0.341\n",
      "Montebello Road : 0.304\n",
      "Mount Vernon Street : 0.344\n",
      "Mozart Street : 0.325\n",
      "Neponset Avenue : 0.329\n",
      "Newbury Street : 0.332\n",
      "North Street : 0.357\n",
      "Norway Street : 0.318\n",
      "Orchardhill Road : 0.33\n",
      "Park Lane : 0.307\n",
      "Parkman Street : 0.349\n",
      "Peterborough Street : 0.333\n",
      "Phillips Street : 0.334\n",
      "Pinckney Street : 0.311\n",
      "Princeton Street : 0.327\n",
      "Quint Avenue : 0.324\n",
      "Revere Street : 0.336\n",
      "Rutland Street : 0.314\n",
      "Sagamore Street : 0.338\n",
      "Saint Botolph Street : 0.285\n",
      "Saint Germain Street : 0.32\n",
      "Salem Street : 0.345\n",
      "Saratoga Street : 0.287\n",
      "Sawyer Avenue : 0.34\n",
      "School Street : 0.316\n",
      "Sentry Hill Place : 0.348\n",
      "Silver Street : 0.343\n",
      "Smith Court : 0.345\n",
      "Soley Street : 0.345\n",
      "South Huntington Avenue : 0.329\n",
      "Strathmore Road : 0.329\n",
      "Sullivan Street : 0.343\n",
      "Sumner Street : 0.353\n",
      "Telegraph Street : 0.353\n",
      "Temple Street : 0.34\n",
      "Thornton Street : 0.349\n",
      "Torpie Street : 0.324\n",
      "Tremont Street : 0.324\n",
      "Trenton Street : 0.354\n",
      "Tudor Street : 0.349\n",
      "Tyler Street : 0.308\n",
      "Union Park : 0.359\n",
      "Upton Street : 0.337\n",
      "Waltham Street : 0.324\n",
      "Washington Street : 0.342\n",
      "Waverly Street : 0.31\n",
      "Webster Street : 0.306\n",
      "Weld Hill Street : 0.331\n",
      "West Canton Street : 0.334\n",
      "West Cedar Street : 0.339\n",
      "West Second Street : 0.322\n",
      "West Springfield Street : 0.32\n",
      "West Third Street : 0.333\n",
      "Winter Street : 0.35\n",
      "Winthrop Street : 0.326\n",
      "Worcester Square : 0.344\n",
      "Worcester Street : 0.306\n",
      "0.379\n"
     ]
    }
   ],
   "source": [
    "pyro.set_rng_seed(10)\n",
    "max_ = 0\n",
    "#Probability of ROI fixing the bedrooms using do operator\n",
    "for i in range(len(labels['street'])):\n",
    "    val = UsingDo('street', i)\n",
    "    print(labels['street'][i],':',val)\n",
    "    if val>max_:\n",
    "        max_ = val\n",
    "print(max_)\n",
    "print()\n",
    "pyro.set_rng_seed(10)\n",
    "max_=0\n",
    "for i in range(len(labels['street'])):\n",
    "    val = UsingConditional('street', i)\n",
    "    print(labels['street'][i],':',val)\n",
    "    if val>max_:\n",
    "        max_ = val\n",
    "print(max_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n",
      "1\n",
      "2\n",
      "3\n",
      "4\n",
      "5\n",
      "6\n",
      "7\n",
      "8\n",
      "9\n",
      "10\n",
      "11\n",
      "12\n",
      "13\n",
      "14\n",
      "15\n",
      "16\n",
      "17\n",
      "18\n",
      "19\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>neighbourhood</th>\n",
       "      <th>bedrooms</th>\n",
       "      <th>property_type</th>\n",
       "      <th>ROI_simple</th>\n",
       "      <th>ROI_complex</th>\n",
       "      <th>causal_effect</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Allston</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>0.325</td>\n",
       "      <td>0.329</td>\n",
       "      <td>0.004</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Allston</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>Condominium</td>\n",
       "      <td>0.325</td>\n",
       "      <td>0.320</td>\n",
       "      <td>0.005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Allston</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>0.325</td>\n",
       "      <td>0.333</td>\n",
       "      <td>0.008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Allston</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>Condominium</td>\n",
       "      <td>0.325</td>\n",
       "      <td>0.349</td>\n",
       "      <td>0.024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Allston</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>0.325</td>\n",
       "      <td>0.314</td>\n",
       "      <td>0.011</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115</th>\n",
       "      <td>West Roxbury</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>Condominium</td>\n",
       "      <td>0.357</td>\n",
       "      <td>0.314</td>\n",
       "      <td>0.043</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>116</th>\n",
       "      <td>West Roxbury</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>0.357</td>\n",
       "      <td>0.321</td>\n",
       "      <td>0.036</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>117</th>\n",
       "      <td>West Roxbury</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>Condominium</td>\n",
       "      <td>0.357</td>\n",
       "      <td>0.339</td>\n",
       "      <td>0.018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>118</th>\n",
       "      <td>West Roxbury</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>0.357</td>\n",
       "      <td>0.297</td>\n",
       "      <td>0.060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>119</th>\n",
       "      <td>West Roxbury</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>Condominium</td>\n",
       "      <td>0.357</td>\n",
       "      <td>0.346</td>\n",
       "      <td>0.011</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>120 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    neighbourhood bedrooms property_type  ROI_simple  ROI_complex  \\\n",
       "0         Allston    (1,2]     Apartment       0.325        0.329   \n",
       "1         Allston    (1,2]   Condominium       0.325        0.320   \n",
       "2         Allston    (2,6]     Apartment       0.325        0.333   \n",
       "3         Allston    (2,6]   Condominium       0.325        0.349   \n",
       "4         Allston    [0,1]     Apartment       0.325        0.314   \n",
       "..            ...      ...           ...         ...          ...   \n",
       "115  West Roxbury    (1,2]   Condominium       0.357        0.314   \n",
       "116  West Roxbury    (2,6]     Apartment       0.357        0.321   \n",
       "117  West Roxbury    (2,6]   Condominium       0.357        0.339   \n",
       "118  West Roxbury    [0,1]     Apartment       0.357        0.297   \n",
       "119  West Roxbury    [0,1]   Condominium       0.357        0.346   \n",
       "\n",
       "     causal_effect  \n",
       "0            0.004  \n",
       "1            0.005  \n",
       "2            0.008  \n",
       "3            0.024  \n",
       "4            0.011  \n",
       "..             ...  \n",
       "115          0.043  \n",
       "116          0.036  \n",
       "117          0.018  \n",
       "118          0.060  \n",
       "119          0.011  \n",
       "\n",
       "[120 rows x 6 columns]"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_df = pd.DataFrame(columns=['neighbourhood','bedrooms','property_type','ROI_simple','ROI_complex','causal_effect'])\n",
    "ind = 0\n",
    "for i in range(len(params_neighbourhood)):\n",
    "    print(i)\n",
    "    intervention_model_simple = pyro.do(pyroModel, data={'neighbourhood': torch.tensor(i)} )\n",
    "    posterior_simple = pyro.infer.Importance(intervention_model_simple, num_samples=5000).run()\n",
    "    marginal_simple = EmpiricalMarginal(posterior_simple,\"roi\")\n",
    "    samples_simple = [marginal_simple().item() for _ in range(1000)]\n",
    "    unique, counts = np.unique(samples_simple, return_counts=True)\n",
    "    roi1 = counts[1]/sum(counts)\n",
    "    for j in range(len(params_bedrooms)):\n",
    "        for k in range(len(params_property_type)):\n",
    "#             print(\"i = \",i,\", j = \",j,\", k = \",k)\n",
    "            \n",
    "            \n",
    "            \n",
    "            intervention_model_complex = pyro.do(pyroModel, data={'neighbourhood': torch.tensor(i),'property_type':torch.tensor(k),\\\n",
    "                                                                'bedrooms': torch.tensor(j)} )\n",
    "            posterior_complex = pyro.infer.Importance(intervention_model_complex, num_samples=5000).run()\n",
    "            marginal_complex = EmpiricalMarginal(posterior_complex,\"roi\")\n",
    "            samples_complex = [marginal_complex().item() for _ in range(1000)]\n",
    "            unique, counts = np.unique(samples_complex, return_counts=True)\n",
    "            roi2 = counts[1]/sum(counts)\n",
    "\n",
    "            new_df.loc[ind] = [labels['neighbourhood'][i], labels['bedrooms'][j], labels['property_type'][k] , roi1, roi2, abs(roi1-roi2)]\n",
    "            ind+=1\n",
    "\n",
    "\n",
    "new_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    neighbourhood bedrooms property_type  ROI_simple  ROI_complex  \\\n",
      "96        Roxbury    (1,2]     Apartment       0.369        0.303   \n",
      "97        Roxbury    (1,2]   Condominium       0.369        0.344   \n",
      "98        Roxbury    (2,6]     Apartment       0.369        0.353   \n",
      "99        Roxbury    (2,6]   Condominium       0.369        0.328   \n",
      "100       Roxbury    [0,1]     Apartment       0.369        0.349   \n",
      "101       Roxbury    [0,1]   Condominium       0.369        0.324   \n",
      "\n",
      "     causal_effect  \n",
      "96           0.066  \n",
      "97           0.025  \n",
      "98           0.016  \n",
      "99           0.041  \n",
      "100          0.020  \n",
      "101          0.045  \n"
     ]
    }
   ],
   "source": [
    "print(new_df[new_df.ROI_simple == new_df.ROI_simple.max()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   neighbourhood bedrooms property_type  ROI_simple  ROI_complex  \\\n",
      "46    Dorchester    [0,1]     Apartment       0.341        0.375   \n",
      "\n",
      "    causal_effect  \n",
      "46          0.034  \n"
     ]
    }
   ],
   "source": [
    "print(new_df[new_df.ROI_complex == new_df.ROI_complex.max()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   neighbourhood bedrooms property_type  ROI_simple  ROI_complex  \\\n",
      "96       Roxbury    (1,2]     Apartment       0.369        0.303   \n",
      "\n",
      "    causal_effect  \n",
      "96          0.066  \n"
     ]
    }
   ],
   "source": [
    "print(new_df[new_df.causal_effect == new_df.causal_effect.max()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [],
   "source": [
    "roisimple = list(new_df[new_df.ROI_simple == new_df.ROI_simple.max()]['ROI_simple'])\n",
    "roicomplex = list(new_df[new_df.ROI_simple == new_df.ROI_simple.max()]['ROI_complex'])\n",
    "causaleffect = roi"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<BarContainer object of 6 artists>"
      ]
     },
     "execution_count": 105,
     "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": [
    "X = np.arange(6)\n",
    "fig = plt.figure()\n",
    "ax = fig.add_axes([0,0,1,1])\n",
    "ax.bar(X + 0.00, roisimple, color = 'b', width = 0.25)\n",
    "ax.bar(X + 0.25, roicomplex, color = 'g', width = 0.25)\n",
    "# ax.set_xticks(X,('(1,2]', '(1,2]', '(2,6]', '(2,6]', '[0,1]', '[0,1]'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<BarContainer object of 6 artists>"
      ]
     },
     "execution_count": 103,
     "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": [
    "X = np.arange(6)\n",
    "fig = plt.figure()\n",
    "ax = fig.add_axes([0,0,1,1])\n",
    "ax.bar(X + 0.00, causaleffectsaleffect, color = 'b', width = 0.25)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<BarContainer object of 2 artists>"
      ]
     },
     "execution_count": 80,
     "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": [
    "plt.bar(x = ['P(ROI|bedrooms=z))','P(ROI|do(neighborhood=x, property_type=y, bedrooms=z))'],height = [0.368,0.378])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "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>bedrooms</th>\n",
       "      <th>simple_ROI</th>\n",
       "      <th>complex_ROI</th>\n",
       "      <th>Causal Effect</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>(1,2]</td>\n",
       "      <td>0.323</td>\n",
       "      <td>0.326</td>\n",
       "      <td>0.003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(2,6]</td>\n",
       "      <td>0.326</td>\n",
       "      <td>0.325</td>\n",
       "      <td>0.001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>[0,1]</td>\n",
       "      <td>0.341</td>\n",
       "      <td>0.318</td>\n",
       "      <td>0.023</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  bedrooms  simple_ROI  complex_ROI  Causal Effect\n",
       "0    (1,2]       0.323        0.326          0.003\n",
       "1    (2,6]       0.326        0.325          0.001\n",
       "2    [0,1]       0.341        0.318          0.023"
      ]
     },
     "execution_count": 106,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "pyro.set_rng_seed(10)\n",
    "new_df2 = pd.DataFrame(columns=['bedrooms', 'simple_ROI','complex_ROI','Causal Effect'])\n",
    "for i in range(len(params_bedrooms)):\n",
    "    intervention_model_simple = pyro.do(pyroModel, data={'bedrooms': torch.tensor(i)})\n",
    "    posterior_simple = pyro.infer.Importance(intervention_model_simple, num_samples=5000).run()\n",
    "    marginal_simple = EmpiricalMarginal(posterior_simple,\"roi\")\n",
    "    samples_simple = [marginal_simple().item() for _ in range(1000)]\n",
    "    unique, counts = np.unique(samples_simple, return_counts=True)\n",
    "    roi_neigh1 = counts[1]/sum(counts)\n",
    "\n",
    "\n",
    "    intervention_model_complex = pyro.do(pyroModel, data={'neighbourhood': torch.tensor(3),'amenities_score':torch.tensor(1),\\\n",
    "                                                                'mobility':torch.tensor(4) , \\\n",
    "                                                                'publicSchoolScores':torch.tensor(0),'property_type':torch.tensor(0),\\\n",
    "                                                                'closenessToDowntown': torch.tensor(1), 'bedrooms': torch.tensor(i),\\\n",
    "                                                                'bathrooms': torch.tensor(4)} )\n",
    "    posterior_complex = pyro.infer.Importance(intervention_model_complex, num_samples=5000).run()\n",
    "    marginal_complex = EmpiricalMarginal(posterior_complex,\"roi\")\n",
    "    samples_complex = [marginal_complex().item() for _ in range(1000)]\n",
    "    unique, counts = np.unique(samples_complex, return_counts=True)\n",
    "    roi_neigh2 = counts[1]/sum(counts)\n",
    "\n",
    "    new_df2.loc[i] = [labels['bedrooms'][i] , roi_neigh1, roi_neigh2, abs(roi_neigh2-roi_neigh1)]\n",
    "new_df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [],
   "source": [
    "roisimple = list(new_df2['simple_ROI'])\n",
    "roicomplex = list(new_df2['complex_ROI'])\n",
    "causaleffect = list(new_df2['Causal Effect'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<BarContainer object of 3 artists>"
      ]
     },
     "execution_count": 117,
     "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": [
    "X = np.arange(3)\n",
    "fig = plt.figure()\n",
    "ax = fig.add_axes([0,0,1,1])\n",
    "ax.bar(X + 0.00, causaleffect, color = 'b', width = 0.25)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "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>street</th>\n",
       "      <th>bedrooms</th>\n",
       "      <th>property_type</th>\n",
       "      <th>ROI</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Alameda Road</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>0.323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Alameda Road</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>Condominium</td>\n",
       "      <td>0.329</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Alameda Road</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>0.326</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Alameda Road</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>Condominium</td>\n",
       "      <td>0.325</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Alameda Road</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>0.341</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115</th>\n",
       "      <td>Brookline Avenue</td>\n",
       "      <td>(1,2]</td>\n",
       "      <td>Condominium</td>\n",
       "      <td>0.338</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>116</th>\n",
       "      <td>Brookline Avenue</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>0.334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>117</th>\n",
       "      <td>Brookline Avenue</td>\n",
       "      <td>(2,6]</td>\n",
       "      <td>Condominium</td>\n",
       "      <td>0.363</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>118</th>\n",
       "      <td>Brookline Avenue</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>Apartment</td>\n",
       "      <td>0.301</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>119</th>\n",
       "      <td>Brookline Avenue</td>\n",
       "      <td>[0,1]</td>\n",
       "      <td>Condominium</td>\n",
       "      <td>0.335</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>120 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "               street bedrooms property_type    ROI\n",
       "0        Alameda Road    (1,2]     Apartment  0.323\n",
       "1        Alameda Road    (1,2]   Condominium  0.329\n",
       "2        Alameda Road    (2,6]     Apartment  0.326\n",
       "3        Alameda Road    (2,6]   Condominium  0.325\n",
       "4        Alameda Road    [0,1]     Apartment  0.341\n",
       "..                ...      ...           ...    ...\n",
       "115  Brookline Avenue    (1,2]   Condominium  0.338\n",
       "116  Brookline Avenue    (2,6]     Apartment  0.334\n",
       "117  Brookline Avenue    (2,6]   Condominium  0.363\n",
       "118  Brookline Avenue    [0,1]     Apartment  0.301\n",
       "119  Brookline Avenue    [0,1]   Condominium  0.335\n",
       "\n",
       "[120 rows x 4 columns]"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_df = pd.DataFrame(columns=['street','bedrooms','property_type','ROI'])\n",
    "ind = 0\n",
    "for i in range(len(params_street)):\n",
    "    for j in range(len(params_bedrooms)):\n",
    "        for k in range(len(params_property_type)):\n",
    "#             print(\"i = \",i,\", j = \",j,\", k = \",k)\n",
    "            intervention_model_complex = pyro.do(pyroModel, data={'street': torch.tensor(i),'amenities_score':torch.tensor(1),\\\n",
    "                                                                'mobility':torch.tensor(4) , \\\n",
    "                                                                'publicSchoolScores':torch.tensor(0),'property_type':torch.tensor(k),\\\n",
    "                                                                'closenessToDowntown': torch.tensor(1), 'bedrooms': torch.tensor(j),\\\n",
    "                                                                'bathrooms': torch.tensor(4)} )\n",
    "            posterior_complex = pyro.infer.Importance(intervention_model_complex, num_samples=5000).run()\n",
    "            marginal_complex = EmpiricalMarginal(posterior_complex,\"roi\")\n",
    "            samples_complex = [marginal_complex().item() for _ in range(1000)]\n",
    "            unique, counts = np.unique(samples_complex, return_counts=True)\n",
    "            roi = counts[1]/sum(counts)\n",
    "\n",
    "            new_df.loc[ind] = [labels['street'][i], labels['bedrooms'][j], labels['property_type'][k] , roi]\n",
    "            ind+=1\n",
    "\n",
    "\n",
    "new_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              street bedrooms property_type    ROI\n",
      "18  Armington Street    (1,2]     Apartment  0.378\n"
     ]
    }
   ],
   "source": [
    "print(new_df[new_df.ROI == new_df.ROI.max()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Beacon Hill'"
      ]
     },
     "execution_count": 118,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels['neighbourhood'][3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:Airbnb_Analysis] *",
   "language": "python",
   "name": "conda-env-Airbnb_Analysis-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}