hotel_booking.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Big Data Analytics of Hotel Bookings"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data Import"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import modules\n",
    "import pandas as pd\n",
    "from pyspark.sql import SparkSession"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# build spark session and spark context\n",
    "spark = SparkSession.builder \\\n",
    "        .appName(\"hotel\") \\\n",
    "        .getOrCreate()\n",
    "sc = spark.sparkContext"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[Row(hotel='Resort Hotel', is_canceled=0, lead_time=342, arrival_date_year=2015, arrival_date_month='July', arrival_date_week_number=27, arrival_date_day_of_month=1, stays_in_weekend_nights=0, stays_in_week_nights=0, adults=2, children='0', babies=0, meal='BB', country='PRT', market_segment='Direct', distribution_channel='Direct', is_repeated_guest=0, previous_cancellations=0, previous_bookings_not_canceled=0, reserved_room_type='C', assigned_room_type='C', booking_changes=3, deposit_type='No Deposit', agent='NULL', company='NULL', days_in_waiting_list=0, customer_type='Transient', adr=0.0, required_car_parking_spaces=0, total_of_special_requests=0, reservation_status='Check-Out', reservation_status_date='2015-07-01'),\n",
       " Row(hotel='Resort Hotel', is_canceled=0, lead_time=737, arrival_date_year=2015, arrival_date_month='July', arrival_date_week_number=27, arrival_date_day_of_month=1, stays_in_weekend_nights=0, stays_in_week_nights=0, adults=2, children='0', babies=0, meal='BB', country='PRT', market_segment='Direct', distribution_channel='Direct', is_repeated_guest=0, previous_cancellations=0, previous_bookings_not_canceled=0, reserved_room_type='C', assigned_room_type='C', booking_changes=4, deposit_type='No Deposit', agent='NULL', company='NULL', days_in_waiting_list=0, customer_type='Transient', adr=0.0, required_car_parking_spaces=0, total_of_special_requests=0, reservation_status='Check-Out', reservation_status_date='2015-07-01')]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = spark.read.csv('hotel_bookings.csv',  inferSchema=True, header = True)\n",
    "df.take(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- hotel: string (nullable = true)\n",
      " |-- is_canceled: integer (nullable = true)\n",
      " |-- lead_time: integer (nullable = true)\n",
      " |-- arrival_date_year: integer (nullable = true)\n",
      " |-- arrival_date_month: string (nullable = true)\n",
      " |-- arrival_date_week_number: integer (nullable = true)\n",
      " |-- arrival_date_day_of_month: integer (nullable = true)\n",
      " |-- stays_in_weekend_nights: integer (nullable = true)\n",
      " |-- stays_in_week_nights: integer (nullable = true)\n",
      " |-- adults: integer (nullable = true)\n",
      " |-- children: string (nullable = true)\n",
      " |-- babies: integer (nullable = true)\n",
      " |-- meal: string (nullable = true)\n",
      " |-- country: string (nullable = true)\n",
      " |-- market_segment: string (nullable = true)\n",
      " |-- distribution_channel: string (nullable = true)\n",
      " |-- is_repeated_guest: integer (nullable = true)\n",
      " |-- previous_cancellations: integer (nullable = true)\n",
      " |-- previous_bookings_not_canceled: integer (nullable = true)\n",
      " |-- reserved_room_type: string (nullable = true)\n",
      " |-- assigned_room_type: string (nullable = true)\n",
      " |-- booking_changes: integer (nullable = true)\n",
      " |-- deposit_type: string (nullable = true)\n",
      " |-- agent: string (nullable = true)\n",
      " |-- company: string (nullable = true)\n",
      " |-- days_in_waiting_list: integer (nullable = true)\n",
      " |-- customer_type: string (nullable = true)\n",
      " |-- adr: double (nullable = true)\n",
      " |-- required_car_parking_spaces: integer (nullable = true)\n",
      " |-- total_of_special_requests: integer (nullable = true)\n",
      " |-- reservation_status: string (nullable = true)\n",
      " |-- reservation_status_date: string (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# view columns and schema\n",
    "df.printSchema()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----+-----------+---------+-----------------+------------------+------------------------+-------------------------+-----------------------+--------------------+------+--------+------+----+-------+--------------+--------------------+-----------------+----------------------+------------------------------+------------------+------------------+---------------+------------+-----+-------+--------------------+-------------+---+---------------------------+-------------------------+------------------+-----------------------+\n",
      "|hotel|is_canceled|lead_time|arrival_date_year|arrival_date_month|arrival_date_week_number|arrival_date_day_of_month|stays_in_weekend_nights|stays_in_week_nights|adults|children|babies|meal|country|market_segment|distribution_channel|is_repeated_guest|previous_cancellations|previous_bookings_not_canceled|reserved_room_type|assigned_room_type|booking_changes|deposit_type|agent|company|days_in_waiting_list|customer_type|adr|required_car_parking_spaces|total_of_special_requests|reservation_status|reservation_status_date|\n",
      "+-----+-----------+---------+-----------------+------------------+------------------------+-------------------------+-----------------------+--------------------+------+--------+------+----+-------+--------------+--------------------+-----------------+----------------------+------------------------------+------------------+------------------+---------------+------------+-----+-------+--------------------+-------------+---+---------------------------+-------------------------+------------------+-----------------------+\n",
      "|    0|          0|        0|                0|                 0|                       0|                        0|                      0|                   0|     0|       4|     0|   0|    488|             0|                   0|                0|                     0|                             0|                 0|                 0|              0|           0|16340| 112593|                   0|            0|  0|                          0|                        0|                 0|                      0|\n",
      "+-----+-----------+---------+-----------------+------------------+------------------------+-------------------------+-----------------------+--------------------+------+--------+------+----+-------+--------------+--------------------+-----------------+----------------------+------------------------------+------------------+------------------+---------------+------------+-----+-------+--------------------+-------------+---+---------------------------+-------------------------+------------------+-----------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# replace the strings \"NULL\" and \"NA\" with null value\n",
    "df_withNull = df.replace('NULL', None).replace('NA', None)\n",
    "\n",
    "# view number of null values per column\n",
    "from pyspark.sql.functions import isnan, when, count, col\n",
    "df_withNull.select([count(when(col(c).isNull(), c)).alias(c) for c in df_withNull.columns]).show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# replace null values in 'children' to 0 since there are only 4\n",
    "df2 = df_withNull.fillna({'children':0})\n",
    "\n",
    "# replace 'children' datatype to int\n",
    "df2 = df2.withColumn('children', col('children').cast(\"Int\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# drop 'company' and 'agent' due to high null count\n",
    "df2 = df2.drop('agent', 'company','country', 'arrival_date_week_number')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('hotel', 'string'),\n",
       " ('is_canceled', 'int'),\n",
       " ('lead_time', 'int'),\n",
       " ('arrival_date_year', 'int'),\n",
       " ('arrival_date_month', 'string'),\n",
       " ('arrival_date_day_of_month', 'int'),\n",
       " ('stays_in_weekend_nights', 'int'),\n",
       " ('stays_in_week_nights', 'int'),\n",
       " ('adults', 'int'),\n",
       " ('children', 'int'),\n",
       " ('babies', 'int'),\n",
       " ('meal', 'string'),\n",
       " ('market_segment', 'string'),\n",
       " ('distribution_channel', 'string'),\n",
       " ('is_repeated_guest', 'int'),\n",
       " ('previous_cancellations', 'int'),\n",
       " ('previous_bookings_not_canceled', 'int'),\n",
       " ('reserved_room_type', 'string'),\n",
       " ('assigned_room_type', 'string'),\n",
       " ('booking_changes', 'int'),\n",
       " ('deposit_type', 'string'),\n",
       " ('days_in_waiting_list', 'int'),\n",
       " ('customer_type', 'string'),\n",
       " ('adr', 'double'),\n",
       " ('required_car_parking_spaces', 'int'),\n",
       " ('total_of_special_requests', 'int'),\n",
       " ('reservation_status', 'string'),\n",
       " ('reservation_status_date', 'string')]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------------+-----+\n",
      "|arrival_date_month|count|\n",
      "+------------------+-----+\n",
      "|              July|12661|\n",
      "|          November| 6794|\n",
      "|          February| 8068|\n",
      "|           January| 5929|\n",
      "|             March| 9794|\n",
      "|           October|11160|\n",
      "|               May|11791|\n",
      "|            August|13877|\n",
      "|             April|11089|\n",
      "|              June|10939|\n",
      "|          December| 6780|\n",
      "|         September|10508|\n",
      "+------------------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df2.groupBy('arrival_date_month').count().show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "#month_num = df2.replace('January', '1')\\\n",
    "#                .replace('February', '2')\\\n",
    "#                .replace('March', '3')\\\n",
    "#                .replace('April', '4')\\\n",
    "#                .replace('May', '5')\\\n",
    "#                .replace('June', '6')\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "#month_num = month_num.replace('July', '7')\\\n",
    "#                      .replace('August', '8')\\\n",
    "#                .replace('September', '9')\\\n",
    "#                .replace('October', '10')\\\n",
    "#                .replace('November', '11')\\\n",
    "#                .replace('December', '12')\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Data Preprocessing Todo's**\n",
    "- map arrival_date_month from strings to numbers: so \"January\" = 1, etc.\n",
    "- numerically encode all the other string variables, otherwise remove them  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# numerically encode arrival_date_month\n",
    "#done in next cell?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "#numerically encode all columns of type string\n",
    "\n",
    "from pyspark.ml import Pipeline\n",
    "from pyspark.ml.feature import OneHotEncoder, OneHotEncoderModel, StringIndexer\n",
    "\n",
    "\n",
    "col_string=['hotel', 'arrival_date_month','meal','market_segment','distribution_channel','reserved_room_type','assigned_room_type','deposit_type','customer_type','reservation_status','reservation_status_date']\n",
    "col_num=[x+\"_NUMERIC\" for x in col_string]\n",
    "\n",
    "#add other columns for encoding\n",
    "col_num=col_num+['arrival_date_year','arrival_date_day_of_month']\n",
    "\n",
    "col_oh=[x+\"_oh\" for x in col_string]\n",
    "\n",
    "col_oh=col_oh+['arrival_date_year_oh','arrival_date_day_of_month_oh']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "indexers = [StringIndexer(inputCol=column, outputCol=column+\"_NUMERIC\").fit(df2) for column in col_string]\n",
    "pipeline = Pipeline(stages=indexers)\n",
    "df_indexed = pipeline.fit(df2).transform(df2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('hotel', 'string'),\n",
       " ('is_canceled', 'int'),\n",
       " ('lead_time', 'int'),\n",
       " ('arrival_date_year', 'int'),\n",
       " ('arrival_date_month', 'string'),\n",
       " ('arrival_date_day_of_month', 'int'),\n",
       " ('stays_in_weekend_nights', 'int'),\n",
       " ('stays_in_week_nights', 'int'),\n",
       " ('adults', 'int'),\n",
       " ('children', 'int'),\n",
       " ('babies', 'int'),\n",
       " ('meal', 'string'),\n",
       " ('market_segment', 'string'),\n",
       " ('distribution_channel', 'string'),\n",
       " ('is_repeated_guest', 'int'),\n",
       " ('previous_cancellations', 'int'),\n",
       " ('previous_bookings_not_canceled', 'int'),\n",
       " ('reserved_room_type', 'string'),\n",
       " ('assigned_room_type', 'string'),\n",
       " ('booking_changes', 'int'),\n",
       " ('deposit_type', 'string'),\n",
       " ('days_in_waiting_list', 'int'),\n",
       " ('customer_type', 'string'),\n",
       " ('adr', 'double'),\n",
       " ('required_car_parking_spaces', 'int'),\n",
       " ('total_of_special_requests', 'int'),\n",
       " ('reservation_status', 'string'),\n",
       " ('reservation_status_date', 'string'),\n",
       " ('hotel_NUMERIC', 'double'),\n",
       " ('arrival_date_month_NUMERIC', 'double'),\n",
       " ('meal_NUMERIC', 'double'),\n",
       " ('market_segment_NUMERIC', 'double'),\n",
       " ('distribution_channel_NUMERIC', 'double'),\n",
       " ('reserved_room_type_NUMERIC', 'double'),\n",
       " ('assigned_room_type_NUMERIC', 'double'),\n",
       " ('deposit_type_NUMERIC', 'double'),\n",
       " ('customer_type_NUMERIC', 'double'),\n",
       " ('reservation_status_NUMERIC', 'double'),\n",
       " ('reservation_status_date_NUMERIC', 'double')]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_indexed.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "#one-hot encode all columns in col_num\n",
    "\n",
    "ohe = OneHotEncoder(dropLast=False)\n",
    "ohe.setInputCols(col_num)\n",
    "ohe.setOutputCols(col_oh)\n",
    "model = ohe.fit(df_indexed)\n",
    "\n",
    "df_casted=model.transform(df_indexed)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('hotel', 'string'),\n",
       " ('is_canceled', 'int'),\n",
       " ('lead_time', 'int'),\n",
       " ('arrival_date_year', 'int'),\n",
       " ('arrival_date_month', 'string'),\n",
       " ('arrival_date_day_of_month', 'int'),\n",
       " ('stays_in_weekend_nights', 'int'),\n",
       " ('stays_in_week_nights', 'int'),\n",
       " ('adults', 'int'),\n",
       " ('children', 'int'),\n",
       " ('babies', 'int'),\n",
       " ('meal', 'string'),\n",
       " ('market_segment', 'string'),\n",
       " ('distribution_channel', 'string'),\n",
       " ('is_repeated_guest', 'int'),\n",
       " ('previous_cancellations', 'int'),\n",
       " ('previous_bookings_not_canceled', 'int'),\n",
       " ('reserved_room_type', 'string'),\n",
       " ('assigned_room_type', 'string'),\n",
       " ('booking_changes', 'int'),\n",
       " ('deposit_type', 'string'),\n",
       " ('days_in_waiting_list', 'int'),\n",
       " ('customer_type', 'string'),\n",
       " ('adr', 'double'),\n",
       " ('required_car_parking_spaces', 'int'),\n",
       " ('total_of_special_requests', 'int'),\n",
       " ('reservation_status', 'string'),\n",
       " ('reservation_status_date', 'string'),\n",
       " ('hotel_NUMERIC', 'double'),\n",
       " ('arrival_date_month_NUMERIC', 'double'),\n",
       " ('meal_NUMERIC', 'double'),\n",
       " ('market_segment_NUMERIC', 'double'),\n",
       " ('distribution_channel_NUMERIC', 'double'),\n",
       " ('reserved_room_type_NUMERIC', 'double'),\n",
       " ('assigned_room_type_NUMERIC', 'double'),\n",
       " ('deposit_type_NUMERIC', 'double'),\n",
       " ('customer_type_NUMERIC', 'double'),\n",
       " ('reservation_status_NUMERIC', 'double'),\n",
       " ('reservation_status_date_NUMERIC', 'double'),\n",
       " ('distribution_channel_oh', 'vector'),\n",
       " ('customer_type_oh', 'vector'),\n",
       " ('reservation_status_date_oh', 'vector'),\n",
       " ('arrival_date_day_of_month_oh', 'vector'),\n",
       " ('reservation_status_oh', 'vector'),\n",
       " ('market_segment_oh', 'vector'),\n",
       " ('reserved_room_type_oh', 'vector'),\n",
       " ('assigned_room_type_oh', 'vector'),\n",
       " ('meal_oh', 'vector'),\n",
       " ('hotel_oh', 'vector'),\n",
       " ('deposit_type_oh', 'vector'),\n",
       " ('arrival_date_year_oh', 'vector'),\n",
       " ('arrival_date_month_oh', 'vector')]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_casted.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---------------+------+\n",
      "|booking_changes| count|\n",
      "+---------------+------+\n",
      "|             12|     2|\n",
      "|              1| 12701|\n",
      "|             13|     5|\n",
      "|              6|    63|\n",
      "|             16|     2|\n",
      "|              3|   927|\n",
      "|             20|     1|\n",
      "|              5|   118|\n",
      "|             15|     3|\n",
      "|             17|     2|\n",
      "|              9|     8|\n",
      "|              4|   376|\n",
      "|              8|    17|\n",
      "|              7|    31|\n",
      "|             10|     6|\n",
      "|             21|     1|\n",
      "|             11|     2|\n",
      "|             14|     5|\n",
      "|              2|  3805|\n",
      "|              0|101314|\n",
      "+---------------+------+\n",
      "only showing top 20 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#check oh encoded df_casted\n",
    "df_casted.groupBy('booking_changes').count().show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+--------------------+-----+\n",
      "|arrival_date_year_oh|count|\n",
      "+--------------------+-----+\n",
      "| (2018,[2017],[1.0])|40687|\n",
      "| (2018,[2015],[1.0])|21996|\n",
      "| (2018,[2016],[1.0])|56707|\n",
      "+--------------------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df_casted.groupBy('arrival_date_year_oh').count().show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+----------------------------+-----+\n",
      "|arrival_date_day_of_month_oh|count|\n",
      "+----------------------------+-----+\n",
      "|             (32,[17],[1.0])| 4406|\n",
      "|             (32,[14],[1.0])| 3819|\n",
      "|              (32,[3],[1.0])| 3855|\n",
      "|             (32,[10],[1.0])| 3575|\n",
      "|              (32,[4],[1.0])| 3763|\n",
      "|             (32,[25],[1.0])| 4160|\n",
      "|             (32,[27],[1.0])| 3802|\n",
      "|             (32,[29],[1.0])| 3580|\n",
      "|             (32,[21],[1.0])| 3767|\n",
      "|             (32,[19],[1.0])| 4052|\n",
      "|             (32,[15],[1.0])| 4196|\n",
      "|              (32,[9],[1.0])| 4096|\n",
      "|             (32,[31],[1.0])| 2208|\n",
      "|             (32,[13],[1.0])| 3745|\n",
      "|             (32,[28],[1.0])| 3946|\n",
      "|             (32,[24],[1.0])| 3993|\n",
      "|              (32,[6],[1.0])| 3833|\n",
      "|             (32,[11],[1.0])| 3599|\n",
      "|             (32,[18],[1.0])| 4002|\n",
      "|             (32,[22],[1.0])| 3596|\n",
      "+----------------------------+-----+\n",
      "only showing top 20 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df_casted.groupBy('arrival_date_day_of_month_oh').count().show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_encoded=df_casted.drop(*col_string)\n",
    "df_encoded=df_encoded.drop(*col_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('is_canceled', 'int'),\n",
       " ('lead_time', 'int'),\n",
       " ('stays_in_weekend_nights', 'int'),\n",
       " ('stays_in_week_nights', 'int'),\n",
       " ('adults', 'int'),\n",
       " ('children', 'int'),\n",
       " ('babies', 'int'),\n",
       " ('is_repeated_guest', 'int'),\n",
       " ('previous_cancellations', 'int'),\n",
       " ('previous_bookings_not_canceled', 'int'),\n",
       " ('booking_changes', 'int'),\n",
       " ('days_in_waiting_list', 'int'),\n",
       " ('adr', 'double'),\n",
       " ('required_car_parking_spaces', 'int'),\n",
       " ('total_of_special_requests', 'int'),\n",
       " ('distribution_channel_oh', 'vector'),\n",
       " ('customer_type_oh', 'vector'),\n",
       " ('reservation_status_date_oh', 'vector'),\n",
       " ('arrival_date_day_of_month_oh', 'vector'),\n",
       " ('reservation_status_oh', 'vector'),\n",
       " ('market_segment_oh', 'vector'),\n",
       " ('reserved_room_type_oh', 'vector'),\n",
       " ('assigned_room_type_oh', 'vector'),\n",
       " ('meal_oh', 'vector'),\n",
       " ('hotel_oh', 'vector'),\n",
       " ('deposit_type_oh', 'vector'),\n",
       " ('arrival_date_year_oh', 'vector'),\n",
       " ('arrival_date_month_oh', 'vector')]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_encoded.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---------------------+-----+\n",
      "|arrival_date_month_oh|count|\n",
      "+---------------------+-----+\n",
      "|       (12,[0],[1.0])|13877|\n",
      "|       (12,[9],[1.0])| 6794|\n",
      "|       (12,[8],[1.0])| 8068|\n",
      "|      (12,[10],[1.0])| 6780|\n",
      "|       (12,[1],[1.0])|12661|\n",
      "|       (12,[4],[1.0])|11089|\n",
      "|       (12,[3],[1.0])|11160|\n",
      "|       (12,[6],[1.0])|10508|\n",
      "|       (12,[5],[1.0])|10939|\n",
      "|       (12,[2],[1.0])|11791|\n",
      "|       (12,[7],[1.0])| 9794|\n",
      "|      (12,[11],[1.0])| 5929|\n",
      "+---------------------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df_encoded.groupBy('arrival_date_month_oh').count().show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----------+---------+-----------------------+--------------------+------+--------+------+-----------------+----------------------+------------------------------+---------------+--------------------+---+---------------------------+-------------------------+-----------------------+----------------+--------------------------+----------------------------+---------------------+-----------------+---------------------+---------------------+-------------+-------------+---------------+--------------------+---------------------+\n",
      "|is_canceled|lead_time|stays_in_weekend_nights|stays_in_week_nights|adults|children|babies|is_repeated_guest|previous_cancellations|previous_bookings_not_canceled|booking_changes|days_in_waiting_list|adr|required_car_parking_spaces|total_of_special_requests|distribution_channel_oh|customer_type_oh|reservation_status_date_oh|arrival_date_day_of_month_oh|reservation_status_oh|market_segment_oh|reserved_room_type_oh|assigned_room_type_oh|      meal_oh|     hotel_oh|deposit_type_oh|arrival_date_year_oh|arrival_date_month_oh|\n",
      "+-----------+---------+-----------------------+--------------------+------+--------+------+-----------------+----------------------+------------------------------+---------------+--------------------+---+---------------------------+-------------------------+-----------------------+----------------+--------------------------+----------------------------+---------------------+-----------------+---------------------+---------------------+-------------+-------------+---------------+--------------------+---------------------+\n",
      "|          0|      342|                      0|                   0|     2|       0|     0|                0|                     0|                             0|              3|                   0|0.0|                          0|                        0|          (5,[1],[1.0])|   (4,[0],[1.0])|         (926,[825],[1.0])|              (32,[1],[1.0])|        (3,[0],[1.0])|    (8,[3],[1.0])|       (10,[6],[1.0])|       (12,[5],[1.0])|(5,[0],[1.0])|(2,[1],[1.0])|  (3,[0],[1.0])| (2018,[2015],[1.0])|       (12,[1],[1.0])|\n",
      "|          0|      737|                      0|                   0|     2|       0|     0|                0|                     0|                             0|              4|                   0|0.0|                          0|                        0|          (5,[1],[1.0])|   (4,[0],[1.0])|         (926,[825],[1.0])|              (32,[1],[1.0])|        (3,[0],[1.0])|    (8,[3],[1.0])|       (10,[6],[1.0])|       (12,[5],[1.0])|(5,[0],[1.0])|(2,[1],[1.0])|  (3,[0],[1.0])| (2018,[2015],[1.0])|       (12,[1],[1.0])|\n",
      "+-----------+---------+-----------------------+--------------------+------+--------+------+-----------------+----------------------+------------------------------+---------------+--------------------+---+---------------------------+-------------------------+-----------------------+----------------+--------------------------+----------------------------+---------------------+-----------------+---------------------+---------------------+-------------+-------------+---------------+--------------------+---------------------+\n",
      "only showing top 2 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df_encoded.show(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['is_canceled',\n",
       " 'lead_time',\n",
       " 'stays_in_weekend_nights',\n",
       " 'stays_in_week_nights',\n",
       " 'adults',\n",
       " 'children',\n",
       " 'babies',\n",
       " 'is_repeated_guest',\n",
       " 'previous_cancellations',\n",
       " 'previous_bookings_not_canceled',\n",
       " 'booking_changes',\n",
       " 'days_in_waiting_list',\n",
       " 'adr',\n",
       " 'required_car_parking_spaces',\n",
       " 'total_of_special_requests',\n",
       " 'distribution_channel_oh',\n",
       " 'customer_type_oh',\n",
       " 'reservation_status_date_oh',\n",
       " 'arrival_date_day_of_month_oh',\n",
       " 'reservation_status_oh',\n",
       " 'market_segment_oh',\n",
       " 'reserved_room_type_oh',\n",
       " 'assigned_room_type_oh',\n",
       " 'meal_oh',\n",
       " 'hotel_oh',\n",
       " 'deposit_type_oh',\n",
       " 'arrival_date_year_oh',\n",
       " 'arrival_date_month_oh']"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "columns_ = df_encoded.columns\n",
    "columns_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Normalized using L^1 norm\n"
     ]
    }
   ],
   "source": [
    "from pyspark.ml.feature import Normalizer\n",
    "from pyspark.ml.linalg import Vectors\n",
    "\n",
    "df_encoded=df_casted.drop(*col_string)\n",
    "df_encoded=df_encoded.drop(*col_num)\n",
    "\n",
    "#columns of all vector type data that are going to be normalied \n",
    "columns_ = ['distribution_channel_oh',\n",
    " 'customer_type_oh',\n",
    " 'reservation_status_date_oh',\n",
    " 'arrival_date_day_of_month_oh',\n",
    " 'reservation_status_oh',\n",
    " 'market_segment_oh',\n",
    " 'reserved_room_type_oh',\n",
    " 'assigned_room_type_oh',\n",
    " 'meal_oh',\n",
    " 'hotel_oh',\n",
    " 'deposit_type_oh',\n",
    " 'arrival_date_year_oh',\n",
    " 'arrival_date_month_oh']\n",
    "\n",
    "\n",
    "#normalizing all vector type data with l^1 norm\n",
    "for column in columns_:\n",
    "    normalizer = Normalizer(inputCol=column, outputCol=column+\"_NORM\", p=1.0) \n",
    "    l1NormData = normalizer.transform(df_encoded)\n",
    "    df_encoded = l1NormData\n",
    "l1NormData = df_encoded\n",
    "print(\"Normalized using L^1 norm\")\n",
    "\n",
    "l1NormData = l1NormData.drop(*columns_)\n",
    "#l1NormData.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('is_canceled', 'int'),\n",
       " ('lead_time', 'int'),\n",
       " ('stays_in_weekend_nights', 'int'),\n",
       " ('stays_in_week_nights', 'int'),\n",
       " ('adults', 'int'),\n",
       " ('children', 'int'),\n",
       " ('babies', 'int'),\n",
       " ('is_repeated_guest', 'int'),\n",
       " ('previous_cancellations', 'int'),\n",
       " ('previous_bookings_not_canceled', 'int'),\n",
       " ('booking_changes', 'int'),\n",
       " ('days_in_waiting_list', 'int'),\n",
       " ('adr', 'double'),\n",
       " ('required_car_parking_spaces', 'int'),\n",
       " ('total_of_special_requests', 'int'),\n",
       " ('distribution_channel_oh_NORM', 'vector'),\n",
       " ('customer_type_oh_NORM', 'vector'),\n",
       " ('reservation_status_date_oh_NORM', 'vector'),\n",
       " ('arrival_date_day_of_month_oh_NORM', 'vector'),\n",
       " ('reservation_status_oh_NORM', 'vector'),\n",
       " ('market_segment_oh_NORM', 'vector'),\n",
       " ('reserved_room_type_oh_NORM', 'vector'),\n",
       " ('assigned_room_type_oh_NORM', 'vector'),\n",
       " ('meal_oh_NORM', 'vector'),\n",
       " ('hotel_oh_NORM', 'vector'),\n",
       " ('deposit_type_oh_NORM', 'vector'),\n",
       " ('arrival_date_year_oh_NORM', 'vector'),\n",
       " ('arrival_date_month_oh_NORM', 'vector')]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "l1NormData.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert to RDD\n",
    "# data_rdd = df2.rdd\n",
    "\n",
    "data_rdd = df_encoded.rdd\n",
    "\n",
    "seed = 42\n",
    "\n",
    "# 70/30 test train split\n",
    "train, test = data_rdd.randomSplit([0.7, 0.3], seed)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Exploratory Data Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt \n",
    "from pyspark.sql.functions import *\n",
    "import pyspark.sql.functions as f\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+------------------+-----------------------+--------------------+------------------+\n",
      "|summary|               adr|stays_in_weekend_nights|stays_in_week_nights|         lead_time|\n",
      "+-------+------------------+-----------------------+--------------------+------------------+\n",
      "|  count|            119390|                 119390|              119390|            119390|\n",
      "|   mean|101.83112153446453|     0.9275986263506156|   2.500301532791691|104.01141636652986|\n",
      "| stddev| 50.53579028554864|     0.9986134945978816|  1.9082856150479093|106.86309704798747|\n",
      "|    min|             -6.38|                      0|                   0|                 0|\n",
      "|    max|            5400.0|                     19|                  50|               737|\n",
      "+-------+------------------+-----------------------+--------------------+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#average daily rate: defined by dividing the sum of all lodging transactions by the total number of staying nights\n",
    "#Number of stays in the weekday/weekend\n",
    "#lead_time: Number of days that elapsed between the entering date of the booking into the PMS and the arrival date\n",
    "df2.select(['adr','stays_in_weekend_nights', 'stays_in_week_nights', 'lead_time']).describe().show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+------------------+-------------------+--------------------+\n",
      "|summary|            adults|           children|              babies|\n",
      "+-------+------------------+-------------------+--------------------+\n",
      "|  count|            119390|             119390|              119390|\n",
      "|   mean|1.8564033838679956|0.10388642264846303|0.007948739425412514|\n",
      "| stddev|0.5792609988327523| 0.3985552216824472| 0.09743619130126377|\n",
      "|    min|                 0|                  0|                   0|\n",
      "|    max|                55|                 10|                  10|\n",
      "+-------+------------------+-------------------+--------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#Whether patrons/customers were adults with children\n",
    "df2.select(['adults', 'children', 'babies']).describe().show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------+------------+-----------+------------------+\n",
      "|deposit_type|       hotel|is_canceled|          avg(adr)|\n",
      "+------------+------------+-----------+------------------+\n",
      "|  No Deposit|  City Hotel|          0|105.75264665136447|\n",
      "|  No Deposit|  City Hotel|          1|112.00739824145559|\n",
      "|  No Deposit|Resort Hotel|          0| 90.93364290932067|\n",
      "|  No Deposit|Resort Hotel|          1|112.77001693121677|\n",
      "|  Non Refund|  City Hotel|          0| 99.77833333333335|\n",
      "|  Non Refund|  City Hotel|          1| 93.12289551541632|\n",
      "|  Non Refund|Resort Hotel|          0|  69.8572463768116|\n",
      "|  Non Refund|Resort Hotel|          1| 66.07261818181843|\n",
      "|  Refundable|  City Hotel|          0| 78.04166666666667|\n",
      "|  Refundable|  City Hotel|          1|130.80714285714285|\n",
      "|  Refundable|Resort Hotel|          0| 68.16483333333333|\n",
      "|  Refundable|Resort Hotel|          1| 84.84772727272728|\n",
      "+------------+------------+-----------+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#Indication on if the customer made a deposit to guarantee the booking.\n",
    "#No Deposit – no deposit was made\n",
    "#Non Refund – a deposit was made in the value of the total stay cost\n",
    "#Refundable – a deposit was made with a value under the total cost of stay.\n",
    "\n",
    "df2.groupBy(\"deposit_type\", 'hotel', 'is_canceled').avg('adr').sort(\"deposit_type\", 'hotel', 'is_canceled').show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------+-----------+-----+\n",
      "|deposit_type|is_canceled|count|\n",
      "+------------+-----------+-----+\n",
      "|  No Deposit|          1|29694|\n",
      "|  No Deposit|          0|74947|\n",
      "|  Non Refund|          1|14494|\n",
      "|  Non Refund|          0|   93|\n",
      "|  Refundable|          1|   36|\n",
      "|  Refundable|          0|  126|\n",
      "+------------+-----------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df2.groupBy(\"deposit_type\", 'is_canceled').count().sort(\"deposit_type\").show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------+------------+-----------+-----+\n",
      "|deposit_type|       hotel|is_canceled|count|\n",
      "+------------+------------+-----------+-----+\n",
      "|  No Deposit|  City Hotel|          0|46198|\n",
      "|  No Deposit|  City Hotel|          1|20244|\n",
      "|  No Deposit|Resort Hotel|          0|28749|\n",
      "|  No Deposit|Resort Hotel|          1| 9450|\n",
      "|  Non Refund|  City Hotel|          0|   24|\n",
      "|  Non Refund|  City Hotel|          1|12844|\n",
      "|  Non Refund|Resort Hotel|          0|   69|\n",
      "|  Non Refund|Resort Hotel|          1| 1650|\n",
      "|  Refundable|  City Hotel|          0|    6|\n",
      "|  Refundable|  City Hotel|          1|   14|\n",
      "|  Refundable|Resort Hotel|          0|  120|\n",
      "|  Refundable|Resort Hotel|          1|   22|\n",
      "+------------+------------+-----------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#check count of cancellations for deposit type\n",
    "df2.groupBy(\"deposit_type\", 'hotel', 'is_canceled').count().sort(\"deposit_type\", 'hotel', 'is_canceled').show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It appears that bookings that were made with no deposit, it is more likely for patrons to cancel their bookings. This makes sense as there is no punitive action or monetary loss if they choose to cancel their reservation. This is more likely to also occur for city hotels.\n",
    "\n",
    "For those that required non-refundable deposits, one would think that it makes sense for customers to be making their reservations more conservatively and wisely. Surprisingly, we find that cancellations for these types of bookings occurred more frequently than bookings that provided refundable deposits. When broken down by hotel type, again we see that majority (~10k more) of cancellations were for city hotels. Perhaps there is also influence from whether the trip was business or leisure related (customer type)\n",
    "\n",
    "Lastly, we notice that for deposits that were refundable, the number of cancellations were fairly low. \n",
    "\n",
    "However, the original dataset revealed that about 88% of hotels did not require a deposit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---------------+-----------+-----+\n",
      "|  customer_type|is_canceled|count|\n",
      "+---------------+-----------+-----+\n",
      "|       Contract|          0| 2814|\n",
      "|       Contract|          1| 1262|\n",
      "|          Group|          0|  518|\n",
      "|          Group|          1|   59|\n",
      "|      Transient|          0|53099|\n",
      "|      Transient|          1|36514|\n",
      "|Transient-Party|          0|18735|\n",
      "|Transient-Party|          1| 6389|\n",
      "+---------------+-----------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#Type of booking, assuming one of four categories\n",
    "#Contract - when the booking has an allotment or other type of contract associated to it\n",
    "#Group – when the booking is associated to a group\n",
    "#Transient – when the booking is not part of a group or contract, and is not associated to other transient booking\n",
    "#Transient-party – when the booking is transient, but is associated to at least other transient booking\n",
    "\n",
    "df2.groupBy(\"customer_type\", 'is_canceled').count().sort('customer_type', 'is_canceled').show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['Contract', 'Group', 'Transient', 'Transient-Party']\n"
     ]
    }
   ],
   "source": [
    "x_customer = [val.customer_type for val in df2.select('customer_type').groupBy('customer_type').count().collect()]\n",
    "x_customer.sort()\n",
    "print(x_customer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [],
   "source": [
    "df2_cancel0 = df2.filter(col('is_canceled') == 0)\n",
    "df2_cancel1 = df2.filter(col('is_canceled') == 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---------------+-----+\n",
      "|  customer_type|count|\n",
      "+---------------+-----+\n",
      "|       Contract| 2814|\n",
      "|          Group|  518|\n",
      "|      Transient|53099|\n",
      "|Transient-Party|18735|\n",
      "+---------------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "customer0 = df2_cancel0.groupBy('customer_type').count().sort('customer_type')\n",
    "customer0.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---------------+-----+\n",
      "|  customer_type|count|\n",
      "+---------------+-----+\n",
      "|       Contract| 1262|\n",
      "|          Group|   59|\n",
      "|      Transient|36514|\n",
      "|Transient-Party| 6389|\n",
      "+---------------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "customer1 = df2_cancel1.groupBy('customer_type').count().sort('customer_type')\n",
    "customer1.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[2814, 518, 53099, 18735]"
      ]
     },
     "execution_count": 153,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer0_list = customer0.select(\"count\").rdd.flatMap(lambda x: x).collect()\n",
    "customer0_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[1262, 59, 36514, 6389]"
      ]
     },
     "execution_count": 154,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customer1_list = customer1.select(\"count\").rdd.flatMap(lambda x: x).collect()\n",
    "customer1_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "#visualize customer_type cancellation\n",
    "\n",
    "#Sorted in this order: 'Contract', 'Group', 'Transient', 'Transient-Party'\n",
    "labels = x_customer\n",
    "cancel0_count = customer0_list\n",
    "cancel1_count = customer1_list\n",
    "\n",
    "x = np.arange(len(x_customer))  # the label locations\n",
    "width = 0.35  # the width of the bars\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "rects1 = ax.bar(x - width/2, cancel0_count, width, label='0 = Not Canceled')\n",
    "rects2 = ax.bar(x + width/2, cancel1_count, width, label='1 = Canceled')\n",
    "\n",
    "# Add some text for labels, title and custom x-axis tick labels, etc.\n",
    "ax.set_ylabel('Count')\n",
    "ax.set_title('Count of hotel cancellations by cutomer type')\n",
    "ax.set_xticks(x)\n",
    "ax.set_xticklabels(labels)\n",
    "ax.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the original dataset, 75% of the data were transient while 25% were transient-party"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------------+------------+------------------+\n",
      "|arrival_date_month|       hotel|          avg(adr)|\n",
      "+------------------+------------+------------------+\n",
      "|             April|  City Hotel|111.25183823529417|\n",
      "|             April|Resort Hotel| 77.84949570518152|\n",
      "|            August|  City Hotel|114.68045530446382|\n",
      "|            August|Resort Hotel|186.79057417245596|\n",
      "|          December|  City Hotel| 88.82630687318492|\n",
      "|          December|Resort Hotel| 68.98422960725051|\n",
      "|          February|  City Hotel|   85.088277945619|\n",
      "|          February|Resort Hotel| 55.17193038994517|\n",
      "|           January|  City Hotel| 82.62898554603844|\n",
      "|           January|Resort Hotel|49.461883264933924|\n",
      "|              July|  City Hotel|110.73429154302588|\n",
      "|              July|Resort Hotel|155.18129892849268|\n",
      "|              June|  City Hotel|119.07434127185242|\n",
      "|              June|Resort Hotel|110.44474876847313|\n",
      "|             March|  City Hotel| 92.64311551563952|\n",
      "|             March|Resort Hotel|57.520146882493776|\n",
      "|               May|  City Hotel|121.63856049562627|\n",
      "|               May|Resort Hotel| 78.75813430739032|\n",
      "|          November|  City Hotel| 88.06960064264429|\n",
      "|          November|Resort Hotel|48.273992613869645|\n",
      "+------------------+------------+------------------+\n",
      "only showing top 20 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#Average hotel daily average rates per month by hotel\n",
    "df2.groupBy('arrival_date_month','hotel').avg('adr').sort('arrival_date_month','hotel').show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------------+------------+-----------+-----+\n",
      "|arrival_date_month|       hotel|is_canceled|count|\n",
      "+------------------+------------+-----------+-----+\n",
      "|             April|  City Hotel|          0| 4015|\n",
      "|             April|  City Hotel|          1| 3465|\n",
      "|             April|Resort Hotel|          0| 2550|\n",
      "|             April|Resort Hotel|          1| 1059|\n",
      "|            August|  City Hotel|          0| 5381|\n",
      "|            August|  City Hotel|          1| 3602|\n",
      "|            August|Resort Hotel|          0| 3257|\n",
      "|            August|Resort Hotel|          1| 1637|\n",
      "|          December|  City Hotel|          0| 2392|\n",
      "|          December|  City Hotel|          1| 1740|\n",
      "|          December|Resort Hotel|          0| 2017|\n",
      "|          December|Resort Hotel|          1|  631|\n",
      "|          February|  City Hotel|          0| 3064|\n",
      "|          February|  City Hotel|          1| 1901|\n",
      "|          February|Resort Hotel|          0| 2308|\n",
      "|          February|Resort Hotel|          1|  795|\n",
      "|           January|  City Hotel|          0| 2254|\n",
      "|           January|  City Hotel|          1| 1482|\n",
      "|           January|Resort Hotel|          0| 1868|\n",
      "|           January|Resort Hotel|          1|  325|\n",
      "+------------------+------------+-----------+-----+\n",
      "only showing top 20 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#Hotel cancellation counts by hotel atype and month\n",
    "df2.groupBy('arrival_date_month','hotel', 'is_canceled').count().sort('arrival_date_month','hotel', 'is_canceled').show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['April',\n",
       " 'August',\n",
       " 'December',\n",
       " 'February',\n",
       " 'January',\n",
       " 'July',\n",
       " 'June',\n",
       " 'March',\n",
       " 'May',\n",
       " 'November',\n",
       " 'October',\n",
       " 'September']"
      ]
     },
     "execution_count": 161,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "months_label = [val.arrival_date_month for val in df2.select('arrival_date_month').distinct().collect()] \n",
    "months_label.sort()\n",
    "months_label\n",
    "#unordered original appeared months"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['04', '08', '12', '02', '01', '07', '06', '03', '05', '11', '10', '09']"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#reorder\n",
    "months_label[0] = '04'\n",
    "months_label[1] = '08'\n",
    "months_label[2] = '12'\n",
    "months_label[3] = '02'\n",
    "months_label[4] = '01'\n",
    "months_label[5] = '07'\n",
    "months_label[6] = '06'\n",
    "months_label[7] = '03'\n",
    "months_label[8] = '05'\n",
    "months_label[9] = '11'\n",
    "months_label[10] = '10'\n",
    "months_label[11] = '09'\n",
    "months_label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['01', '02', '03', '04', '05', '06', '07', '08', '09', '10', '11', '12']\n"
     ]
    }
   ],
   "source": [
    "months_label.sort()\n",
    "print(months_label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['January',\n",
       " 'February',\n",
       " 'March',\n",
       " 'April',\n",
       " 'May',\n",
       " 'June',\n",
       " 'July',\n",
       " 'August',\n",
       " 'September',\n",
       " 'October',\n",
       " 'November',\n",
       " 'December']"
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#reorder2\n",
    "months_label[0] = 'January'\n",
    "months_label[1] = 'February'\n",
    "months_label[2] = 'March'\n",
    "months_label[3] = 'April'\n",
    "months_label[4] = 'May'\n",
    "months_label[5] = 'June'\n",
    "months_label[6] = 'July'\n",
    "months_label[7] = 'August'\n",
    "months_label[8] = 'September'\n",
    "months_label[9] = 'October'\n",
    "months_label[10] = 'November'\n",
    "months_label[11] = 'December'\n",
    "months_label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------------+-----+\n",
      "|arrival_date_month|count|\n",
      "+------------------+-----+\n",
      "|             April| 6565|\n",
      "|            August| 8638|\n",
      "|          December| 4409|\n",
      "|          February| 5372|\n",
      "|           January| 4122|\n",
      "|              July| 7919|\n",
      "|              June| 6404|\n",
      "|             March| 6645|\n",
      "|               May| 7114|\n",
      "|          November| 4672|\n",
      "|           October| 6914|\n",
      "|         September| 6392|\n",
      "+------------------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "months_cancel0 = df2_cancel0.groupBy('arrival_date_month').count().sort('arrival_date_month')\n",
    "months_cancel0.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------------+-----+\n",
      "|arrival_date_month|count|\n",
      "+------------------+-----+\n",
      "|             April| 4524|\n",
      "|            August| 5239|\n",
      "|          December| 2371|\n",
      "|          February| 2696|\n",
      "|           January| 1807|\n",
      "|              July| 4742|\n",
      "|              June| 4535|\n",
      "|             March| 3149|\n",
      "|               May| 4677|\n",
      "|          November| 2122|\n",
      "|           October| 4246|\n",
      "|         September| 4116|\n",
      "+------------------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "months_cancel1 = df2_cancel1.groupBy('arrival_date_month').count().sort('arrival_date_month')\n",
    "months_cancel1.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[6565, 8638, 4409, 5372, 4122, 7919, 6404, 6645, 7114, 4672, 6914, 6392]"
      ]
     },
     "execution_count": 158,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "month0_list = months_cancel0.select(\"count\").rdd.flatMap(lambda x: x).collect()\n",
    "month0_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[4524, 5239, 2371, 2696, 1807, 4742, 4535, 3149, 4677, 2122, 4246, 4116]"
      ]
     },
     "execution_count": 160,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "month1_list = months_cancel1.select(\"count\").rdd.flatMap(lambda x: x).collect()\n",
    "month1_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "#visualize non cancellations by month\n",
    "\n",
    "#These are unordered: April, August, December\n",
    "#Only ordered alphabetically\n",
    "labels = months_label\n",
    "cancel0_count = month0_list\n",
    "cancel1_count = month1_list\n",
    "\n",
    "x = np.arange(len(months_label))  # the label locations\n",
    "width = 0.35  # the width of the bars\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "rects1 = ax.bar(x - width/2, cancel0_count, width, label='0 = Not Canceled')\n",
    "rects2 = ax.bar(x + width/2, cancel1_count, width, label='1 = Canceled')\n",
    "\n",
    "# Add some text for labels, title and custom x-axis tick labels, etc.\n",
    "ax.set_ylabel('Count')\n",
    "ax.set_title('Count of hotel cancellations by month')\n",
    "ax.set_xticks(x)\n",
    "ax.set_xticklabels(labels, rotation = 45, ha=\"right\")\n",
    "ax.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----------------+------------+-----------+-----+\n",
      "|arrival_date_year|       hotel|is_canceled|count|\n",
      "+-----------------+------------+-----------+-----+\n",
      "|             2015|  City Hotel|          0| 7678|\n",
      "|             2015|  City Hotel|          1| 6004|\n",
      "|             2015|Resort Hotel|          0| 6176|\n",
      "|             2015|Resort Hotel|          1| 2138|\n",
      "|             2016|  City Hotel|          0|22733|\n",
      "|             2016|  City Hotel|          1|15407|\n",
      "|             2016|Resort Hotel|          0|13637|\n",
      "|             2016|Resort Hotel|          1| 4930|\n",
      "|             2017|  City Hotel|          0|15817|\n",
      "|             2017|  City Hotel|          1|11691|\n",
      "|             2017|Resort Hotel|          0| 9125|\n",
      "|             2017|Resort Hotel|          1| 4054|\n",
      "+-----------------+------------+-----------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#Count of cancellations by hotel type and year\n",
    "#2015-2017\n",
    "df2.groupBy('arrival_date_year','hotel', 'is_canceled').count().sort('arrival_date_year','hotel', 'is_canceled').show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------+-----+\n",
      "|       hotel|count|\n",
      "+------------+-----+\n",
      "|  City Hotel|79330|\n",
      "|Resort Hotel|40060|\n",
      "+------------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#count of cancellation by hotel\n",
    "df2.groupBy('hotel').count().select('hotel', f.col('count')).show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Get counts of cancel or not cancel by hotel type\n",
    "resort_0 = df2.filter((col('hotel') == 'Resort Hotel') & (col('is_canceled') == 0)).count()\n",
    "resort_1 = df2.filter((col('hotel') == 'Resort Hotel') & (col('is_canceled') == 1)).count()\n",
    "city_0 = df2.filter((col('hotel') == 'City Hotel') & (col('is_canceled') == 0)).count()\n",
    "city_1 = df2.filter((col('hotel') == 'City Hotel') & (col('is_canceled') == 1)).count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------+-----------+-----+\n",
      "|       hotel|is_canceled|count|\n",
      "+------------+-----------+-----+\n",
      "|Resort Hotel|          0|28938|\n",
      "|  City Hotel|          0|46228|\n",
      "|Resort Hotel|          1|11122|\n",
      "|  City Hotel|          1|33102|\n",
      "+------------+-----------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#cancellations by hotel type in dataframe\n",
    "df2.groupBy('hotel', 'is_canceled').count().show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "#to visualize, we use matplotlib\n",
    "#Hotel cancellations by hotel type\n",
    "\n",
    "\n",
    "#y_iscancel = [val.is_canceled for val in df2.select('is_canceled').distinct().collect()]   #gets [0,1]\n",
    "x_hoteltype = [val.hotel for val in df2.select('hotel').groupBy('hotel').count().collect()]\n",
    "\n",
    "labels = ['City Hotel', 'Resort Hotel']\n",
    "cancel0_count = [city_0, resort_0]\n",
    "cancel1_count = [city_1, resort_1]\n",
    "\n",
    "x = np.arange(len(x_hoteltype))  # the label locations\n",
    "width = 0.35  # the width of the bars\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "rects1 = ax.bar(x - width/2, cancel0_count, width, label='0 = Not Canceled')\n",
    "rects2 = ax.bar(x + width/2, cancel1_count, width, label='1 = Canceled')\n",
    "\n",
    "# Add some text for labels, title and custom x-axis tick labels, etc.\n",
    "ax.set_ylabel('Count')\n",
    "ax.set_title('Count of hotel cancellations by hotel type')\n",
    "ax.set_xticks(x)\n",
    "ax.set_xticklabels(labels)\n",
    "ax.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------+----+-----+\n",
      "|       hotel|meal|count|\n",
      "+------------+----+-----+\n",
      "|  City Hotel|  BB|62305|\n",
      "|  City Hotel|  FB|   44|\n",
      "|  City Hotel|  HB| 6417|\n",
      "|  City Hotel|  SC|10564|\n",
      "|Resort Hotel|  BB|30005|\n",
      "|Resort Hotel|  FB|  754|\n",
      "|Resort Hotel|  HB| 8046|\n",
      "|Resort Hotel|  SC| 1255|\n",
      "+------------+----+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# for 'meal': \n",
    "#undefined/SC are same, they mean no meal package. \n",
    "#BB: is bed&breakfast\n",
    "#HB: half board, is breakfast and dinner\n",
    "#FB: full board, is full breakfast lunch and dinner\n",
    "\n",
    "# replace Undefined with SC in meal package\n",
    "df2 = df2.replace('Undefined', 'SC')\n",
    "\n",
    "# see value counts for 'meal'\n",
    "df2.groupBy('hotel','meal').count().sort('hotel', 'meal').show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "helpful links:\n",
    "- https://spark.apache.org/docs/latest/ml-guide.html \n",
    "- https://runawayhorse001.github.io/LearningApacheSpark/exploration.html\n",
    "- https://github.com/Bergvca/pyspark_dist_explore\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Modeling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will need to include\n",
    "\n",
    "a) Evaluate metrics\n",
    "- For classificaiton,\n",
    "        - accuracy\n",
    "        - precision, recall, F1 score\n",
    "        - confusion matrix\n",
    "        - area under the ROC curve (AUROC)\n",
    "        \n",
    "b) Sensitivity analysis\n",
    "\n",
    "Sensitivity analysis measures the effect of changing the model inputs or parameters. For example, if\n",
    "the model uses a hyperparameter C, how does AUROC change when feature X is increased/decreased\n",
    "by one standard deviation. The hope is that sensitivity is low."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Benchmark or Simple Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will use logistic regression to predict whether patrons will  cancel their hotel bookings\n",
    "\n",
    "First, we test using logistic regression on just one feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load modules\n",
    "import pandas as pd\n",
    "import pyspark.sql.functions as F\n",
    "import pyspark.mllib.regression as reg\n",
    "from pyspark.mllib.classification import LogisticRegressionWithLBFGS\n",
    "from pyspark.mllib.evaluation import BinaryClassificationMetrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# narrow the list of features for modeling\n",
    "vars_to_keep = ['hotel',\n",
    "                ...]\n",
    "data_rdd2 = data_rdd.select([is_canceled]+vars_to_keep)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "iterations = 10\n",
    "#target = is_canceled"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'data_rdd' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-29-3baaef521655>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# Show some rows to be sure things look ok.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mdata_rdd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m5\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m: name 'data_rdd' is not defined"
     ]
    }
   ],
   "source": [
    "# Show some rows to be sure things look ok.\n",
    "data_rdd.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "#code for defining AUC for classfication (taken from last breast cancer assignment)\n",
    "\n",
    "def compute_univariate_aucs(df, target, training_fraction, iters, seed):\n",
    "\n",
    "    # split the data into train/test using seed\n",
    "    data_train, data_test = df.randomSplit(training_fraction, seed)\n",
    "    \n",
    "    # list of predictor variables\n",
    "    vars = vars_to_keep\n",
    "    \n",
    "    # results storage\n",
    "    df_auc = pd.DataFrame(index=vars, columns=['weight','auroc'])    \n",
    "\n",
    "    for v in vars:    \n",
    "        print('=== analysis of variable: {}'.format(v))\n",
    "\n",
    "        # create train and test dataframes with columns: target, v\n",
    "        datai_tr = data_train.select([target]+[v])\n",
    "        datai_te = data_test.select([target]+[v])\n",
    "\n",
    "        # cast to LabeledPoint\n",
    "        # train\n",
    "        datai_tr_lp = datai_tr \\\n",
    "                     .rdd \\\n",
    "                     .map(lambda row: reg.LabeledPoint(row[0], row[1:]))\n",
    "        \n",
    "        # test\n",
    "        datai_te_lp = datai_te\\\n",
    "                      .rdd\\\n",
    "                      .map(lambda row: reg.LabeledPoint(row[0], row[1:]))\n",
    "\n",
    "        # train logistic regression, setting iterations, including intercept\n",
    "        LR_Model = LogisticRegressionWithLBFGS.train(datai_tr_lp, iterations=iters, intercept=True)\n",
    "\n",
    "        # from test set, zip labels with predicted labels and cast to float\n",
    "        #p.label    # gives label\n",
    "        #p.features # gives features\n",
    "        #comment out because following results say 'p is not defined'\n",
    "        \n",
    "        act_pred_test_set = datai_te_lp.map(lambda p: (p.label, LR_Model.predict(p.features))) \\\n",
    "                                            .map(lambda row: (row[0], row[1] * 1.0))\n",
    "        \n",
    "        metrics = BinaryClassificationMetrics(act_pred_test_set)\n",
    "        # metrics.areaUnderROC\n",
    "        \n",
    "        df_auc['weight'].loc[v] = LR_Model.weights  # store the weights\n",
    "        df_auc['auroc'].loc[v] = metrics.areaUnderROC # extract AUROC\n",
    "        print('=== completed analysis of variable: {}'.format(v))\n",
    "        \n",
    "    df_auc.sort_values(by='auroc', ascending=False, inplace=True)\n",
    "    \n",
    "    return df_auc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#not downsample dataframe using ITERS and SEED that was defined\n",
    "compute_univariate_aucs(data_rdd2, is_canceled, training_fraction, ITERS, SEED)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Champion or Sophisticated Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}