final_hotel_recc.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import ipywidgets as w\n",
    "from IPython.display import display, IFrame\n",
    "import pyspark\n",
    "from pyspark.sql import SQLContext, functions, types\n",
    "from pyspark.sql import Row\n",
    "import matplotlib.pyplot as plt\n",
    "from hotel_recc import *\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "sc=pyspark.SparkContext(appName=\"hotel_recc\")\n",
    "spark = SQLContext(sc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## User details and Preferences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6e82f1f1baec49caa1b1d70a0a7530b2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(VBox(children=(Text(value='', description='User Name'), Text(value='', description='Destination…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "## Ipywidgets to get user name and destination\n",
    "name = w.Text(description=\"User Name\")\n",
    "place = w.Text(description=\"Destination\")\n",
    "v1 = w.VBox([name,place])\n",
    "\n",
    "## Ipywidgets to get start and end date of the trip\n",
    "\n",
    "start = w.DatePicker(description='Start Date',disabled=False)\n",
    "end = w.DatePicker(description='End Date',disabled=False)\n",
    "v2 = w.VBox([start,end])\n",
    "\n",
    "out = w.HBox([v1,v2])\n",
    "display(out)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Reading file containing hotel details after removing duplicates\n",
    "del_dup = spark.read.json('etl/del_dup')\n",
    "\n",
    "## Reading file containing hotel details after removing duplicates and exploding amenities\n",
    "newh_df = spark.read.json('etl/newh_df')\n",
    "\n",
    "del_dup.createOrReplaceTempView('del_dup')\n",
    "newh_df.createOrReplaceTempView('newh_df')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Select atleast 5 amenities:\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dd2033dc45264348a5ac61c05207cf2c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(VBox(children=(Button(description=' Nonsmoking hotel', layout=Layout(height='100px', width='100…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Nonsmoking hotel\n",
      "Select 4 more!\n",
      "\n",
      " Family Rooms\n",
      "Select 3 more!\n",
      "\n",
      " Public Wifi\n",
      "Select 2 more!\n",
      "\n",
      " Pets Allowed  Dog  Pet Friendly \n",
      "Select 1 more!\n",
      "\n",
      " Air conditioning\n"
     ]
    }
   ],
   "source": [
    "## Finding top 15 amentities to ask users to select inorder to provide hotel recommendations based on amenities chosen\n",
    "newh1_df  = spark.sql(\"SELECT amenities,COUNT(amenities) AS tot_count FROM newh_df GROUP BY amenities ORDER BY tot_count DESC\")\n",
    "top_amenities = [x[0] for x in newh1_df.head(16) if x[0] != '']\n",
    "\n",
    "## Getting user's amenity preferences\n",
    "amenities_pref = []\n",
    "def on_button_clicked(b):\n",
    "    if b.description in amenities_pref:\n",
    "        return\n",
    "    print(b.description)\n",
    "    amenities_pref.append(b.description)\n",
    "    if(len(amenities_pref) < 5):\n",
    "        print(\"Select {x} more!\\n\".format(x=5-len(amenities_pref)))\n",
    "        \n",
    "but_layout = w.Layout(width='100%', height='100px')\n",
    "but_items = [w.Button(description=c, layout=but_layout) for c in top_amenities]\n",
    "on_clk = [item.on_click(on_button_clicked) for item in but_items]\n",
    "r1 = w.VBox([i for i in but_items[:5]])\n",
    "r2 = w.VBox([i for i in but_items[5:10]])\n",
    "r3 = w.VBox([i for i in but_items[10:]])\n",
    "buttons = w.HBox([r1,r2,r3])\n",
    "print(\"Select atleast 5 amenities:\")\n",
    "display(buttons)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model training and evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "best rank :  12\n",
      "best RMSE:3.08025294004643\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<BarContainer object of 3 artists>"
      ]
     },
     "execution_count": 7,
     "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": [
    "usr_rating = amenities_rating(spark, amenities_pref, newh_df)\n",
    "rank, error, errors, usrid_s2 = model_train(spark, usr_rating)\n",
    "print(\"best rank : \",rank)\n",
    "print(\"best RMSE:\" +str(error))\n",
    "plt.bar([4,8,12], height=errors)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Getting Recommendation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "u_tempdf = get_hotel_recc(spark, usrid_s2)\n",
    "hotel_df = del_dup.join(u_tempdf, \"id\").withColumn(\"address\",functions.lower(functions.col(\"address\")))\n",
    "user_location = place.value.lower()\n",
    "hotel_sugg = hotel_df.where(hotel_df.address.contains(user_location))\n",
    "recc = hotel_sugg.dropna().toPandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%capture\n",
    "final = dict()\n",
    "final['address'] = recc[:5]['address'].values.tolist()\n",
    "final['amenities'] = recc[:5]['amenities'].values.T.tolist()\n",
    "final['experience'] = recc[:5]['hotel_experience'].values.tolist()\n",
    "final['name'] = recc[:5]['hotel_name'].values.tolist()\n",
    "final['rating'] = recc[:5]['hotel_rating'].values.tolist()\n",
    "final['location'] = [i[1:-1] for i in recc[:5]['location'].values.tolist()]\n",
    "final['price'] = recc[:5]['price'].values.tolist()\n",
    "final['image'] = [get_image(i) for i in recc[:5]['hotel_name'].values.tolist()]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Final Output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a85fa93f117e40819f08a0212079fed2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Tab(children=(VBox(children=(Image(value=b'\\xff\\xd8\\xff\\xe0\\x00\\x10JFIF\\x00\\x01\\x01\\x01\\x00\\x01\\x00\\x01\\x00\\x0…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "days = (end.value - start.value).days\n",
    "tab_recc = get_hotel_output(days, final)\n",
    "display(tab_recc)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}