Meetup4-SparkMadeEasy.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Spark 2.0 made easy - Hands on Lab - USING LINUX DSVM\n",
    "It should take about 11 minutes to run on a LDSVM DS4_v2 of  of 8 cores and 28GB RAM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![SparkAzure Logo](https://msdnshared.blob.core.windows.net/media/2016/08/Spark-Azure.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Using LDSVM\n",
    "\n",
    "## Go to the azure marketplace and click on the GET IT NOW button, and follow the steps.\n",
    "https://azuremarketplace.microsoft.com/en-us/marketplace/apps/microsoft-ads.linux-data-science-vm-ubuntu?tab=Overview\n",
    "\n",
    "A standalone instance of Apache Spark is preinstalled on the Linux DSVM to help you develop Spark applications locally first before testing and deploying on large clusters. You can run PySpark programs through the Jupyter kernel\n",
    "![DSVM_image](https://msdnshared.blob.core.windows.net/media/2017/04/041817_0051_DeepLearnin1.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Introduction to Spark"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "![About Spark](http://curriculum-release.s3-website-us-west-2.amazonaws.com/wiki-book/book_intro/spark_about.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Spark is a unified processing engine that can analyze big data using SQL, machine learning, graph processing or real time stream analysis:\n",
    "\n",
    "![Spark Engines](http://curriculum-release.s3-website-us-west-2.amazonaws.com/wiki-book/book_intro/spark_4engines.png)\n",
    "\n",
    "We will mostly focus on Spark SQL and DataFrames this evening."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Spark can read from many different databases and file systems and run in various environments:\n",
    "\n",
    "![Spark Goal](http://curriculum-release.s3-website-us-west-2.amazonaws.com/wiki-book/book_intro/spark_goal.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Although Spark supports four languages (Scala, Java, Python, R), tonight we will use Python.\n",
    "Broadly speaking, there are **2 APIs** for interacting with Spark:\n",
    "- **DataFrames/SQL/Datasets:** general, higher level API for users of Spark\n",
    "- **RDD:** a lower level API for spark internals and advanced programming"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A Spark cluster is made of one Driver and many Executor JVMs (java virtual machines):"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![Spark Physical Cluster, slots](http://curriculum-release.s3-website-us-west-2.amazonaws.com/wiki-book/book_intro/spark_cluster_slots.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The Driver sends Tasks to the empty slots on the Executors when work has to be done:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![Spark Physical Cluster, tasks](http://curriculum-release.s3-website-us-west-2.amazonaws.com/wiki-book/book_intro/spark_cluster_tasks.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "On Azure Data Science VM, Spark is installed in local mode cluster, where the Driver and Executor code run in the same JVM. Local mode clusters are typically used for prototyping and learning Spark"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Open the Apache Spark 2.0 related documentation in new tabs, so you can easily reference the API guide:\n",
    "\n",
    "1) Spark 2.0.1  docs: https://spark.apache.org/docs/2.0.1/\n",
    "\n",
    "2) DataFrame user documentation: https://spark.apache.org/docs/2.0.1/sql-programming-guide.html\n",
    "\n",
    "3) PySpark API 2.0 docs: http://spark.apache.org/docs/latest/api/python/index.html\n",
    "\n",
    "Other usefull links:\n",
    "\n",
    "4) IPython magic commands: http://ipython.readthedocs.io/en/stable/interactive/magics.html\n",
    "\n",
    "5) Guide for DataFrame Operations in PySpark: https://www.analyticsvidhya.com/blog/2016/10/spark-dataframe-and-operations/\n",
    "\n",
    "6) Repo for Azure Machine Learning- Data Science: https://github.com/Azure/Azure-MachineLearning-DataScience\n",
    "\n",
    "7) Understanding Resource Allocation configurations for Spark apps: http://site.clairvoyantsoft.com/understanding-resource-allocation-configurations-spark-application/\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Spark cluster configuration"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### (OPTIONAL STEP) IF RUNNING THIS NOTEBOOK IN AZURE DATA SCIENCE VM:\n",
    "#### ssh into your machine\n",
    "ssh username@ip<br>\n",
    "sudo locate spark-defaults.conf<br>\n",
    "sudo vi /dsvm/tools/spark/spark-2.0.2/conf/spark-defaults.conf<br>\n",
    "\n",
    "#### then change the line below to fit your VM RAM (use 90% of the available RAM)\n",
    "spark.driver.memory Xg"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's make sure that we have our desired ram"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "u'25g'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sc._conf.get('spark.driver.memory')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### We need now to enable the SparkUI\n",
    "Go to Your Azure portal, click on the Resource Group you just created (where the DSVM is) and click on the Network Security Group Resource and add and Inbound rule for the Spark UI port 4040. It should look like this:\n",
    "\n",
    "Priotity: 1040, Name: SparkUI, Source: Any, Destination: Any, Service: Custom (Any/4040), Action: Allow\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Introduction to Car Rental Bookings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pyspark"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's read or source file - This files contains many columns related to one day of all the diferent car rental companies and their bookings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 16 ms, sys: 0 ns, total: 16 ms\n",
      "Wall time: 52.6 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pricingDF = spark.read.csv(\"wasb://meetup-files@mdbdsmeetup.blob.core.windows.net/COMPETITOR_RATE_20170216.csv\", header=False, inferSchema=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It took about ~50 seconds to read"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Inferring the schema works for ad hoc analysis against smaller datasets. But when working on multi-TB+ data, it's better to provide an **explicit pre-defined schema manually**, so there's no inferring cost:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from pyspark.sql.types import StructType, StructField, IntegerType, StringType, BooleanType, DoubleType"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Note that we are removing all space characters from the col names to prevent errors when writing to Parquet later\n",
    "\n",
    "PricingSchema = StructType([StructField('UniqueIdentity', StringType(), True),\n",
    "                     StructField('PickupLocationCode', StringType(), True),\n",
    "                     StructField('ReturnLocationCode', StringType(), True),\n",
    "                     StructField('CarClass', StringType(), True),\n",
    "                     StructField('PickupDate', StringType(), True),       \n",
    "                     StructField('ReturnDate', StringType(), True),       \n",
    "                     StructField('PricingSegmentId', StringType(), True),       \n",
    "                     StructField('ShopDate', StringType(), True),       \n",
    "                     StructField('CompetitorId', StringType(), True),       \n",
    "                     StructField('LengthOfRent', IntegerType(), True),       \n",
    "                     StructField('PlanCode', StringType(), True),       \n",
    "                     StructField('CurrencyCode', StringType(), True),                  \n",
    "                     StructField('ShopSourceCode', StringType(), True),       \n",
    "                     StructField('ShopVendorCode', StringType(), True),       \n",
    "                     StructField('Rate', DoubleType(), True),       \n",
    "                     StructField('TimeAndMileage', DoubleType(), True),       \n",
    "                     StructField('EstimatedTotal', DoubleType(), True),       \n",
    "                     StructField('LocationIndicator', StringType(), True),       \n",
    "                     StructField('MileageAllowance', StringType(), True),                 \n",
    "                     StructField('CreatedDate', StringType(), True),       \n",
    "                     StructField('EndPickupDate', StringType(), True),       \n",
    "                     StructField('PointOfSaleCountryCode', StringType(), True),       \n",
    "                     StructField('CategoryCode', StringType(), True),       \n",
    "                     StructField('RealTimeCanvasCode', StringType(), True)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 0 ns, sys: 4 ms, total: 4 ms\n",
      "Wall time: 78.7 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pricingDF = spark.read.csv(\"wasb://meetup-files@mdbdsmeetup.blob.core.windows.net/COMPETITOR_RATE_20170216.csv\", header=False, schema=PricingSchema)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['UniqueIdentity',\n",
       " 'PickupLocationCode',\n",
       " 'ReturnLocationCode',\n",
       " 'CarClass',\n",
       " 'PickupDate',\n",
       " 'ReturnDate',\n",
       " 'PricingSegmentId',\n",
       " 'ShopDate',\n",
       " 'CompetitorId',\n",
       " 'LengthOfRent',\n",
       " 'PlanCode',\n",
       " 'CurrencyCode',\n",
       " 'ShopSourceCode',\n",
       " 'ShopVendorCode',\n",
       " 'Rate',\n",
       " 'TimeAndMileage',\n",
       " 'EstimatedTotal',\n",
       " 'LocationIndicator',\n",
       " 'MileageAllowance',\n",
       " 'CreatedDate',\n",
       " 'EndPickupDate',\n",
       " 'PointOfSaleCountryCode',\n",
       " 'CategoryCode',\n",
       " 'RealTimeCanvasCode']"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pricingDF.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 8 ms, sys: 0 ns, total: 8 ms\n",
      "Wall time: 30.2 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "25354815"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "pricingDF.count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Because we gave it the schema, it took less."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+--------------+------------------+------------------+--------+-------------------+-------------------+----------------+-------------------+------------+------------+--------+------------+--------------+--------------+------+--------------+--------------+-----------------+----------------+-------------------+-------------------+----------------------+------------+------------------+\n",
      "|UniqueIdentity|PickupLocationCode|ReturnLocationCode|CarClass|         PickupDate|         ReturnDate|PricingSegmentId|           ShopDate|CompetitorId|LengthOfRent|PlanCode|CurrencyCode|ShopSourceCode|ShopVendorCode|  Rate|TimeAndMileage|EstimatedTotal|LocationIndicator|MileageAllowance|        CreatedDate|      EndPickupDate|PointOfSaleCountryCode|CategoryCode|RealTimeCanvasCode|\n",
      "+--------------+------------------+------------------+--------+-------------------+-------------------+----------------+-------------------+------------+------------+--------+------------+--------------+--------------+------+--------------+--------------+-----------------+----------------+-------------------+-------------------+----------------------+------------+------------------+\n",
      "|   14214956668|               EWR|               EWR|       I|03/03/2017 10:00:00|03/05/2017 10:00:00|              NP|02/16/2017 01:03:03|          ZD|           2|       E|         USD|            GD|            RB|3400.0|        6800.0|       10449.0|                O|             UNL|02/16/2017 00:32:46|03/03/2017 10:00:00|                    US|           A|                 C|\n",
      "|   14214956672|               EWR|               EWR|       S|02/17/2017 10:00:00|02/19/2017 10:00:00|              NP|02/16/2017 01:03:02|          ZI|           2|       E|         USD|            GD|            RB|8370.0|       16740.0|       23062.0|                O|             UNL|02/16/2017 00:32:46|02/17/2017 10:00:00|                    US|           A|                 C|\n",
      "|   14214956676|               EWR|               EWR|       S|02/24/2017 10:00:00|02/26/2017 10:00:00|              NP|02/16/2017 01:03:03|          ET|           2|       E|         USD|            GD|            RB|5615.0|       11230.0|       15629.0|                O|             UNL|02/16/2017 00:32:46|02/24/2017 10:00:00|                    US|           A|                 C|\n",
      "|   14214956679|               EWR|               EWR|       S|03/10/2017 10:00:00|03/12/2017 10:00:00|              NP|02/16/2017 01:03:03|          ZD|           2|       E|         USD|            GD|            RB|3700.0|        7400.0|       11211.0|                O|             UNL|02/16/2017 00:32:46|03/10/2017 10:00:00|                    US|           A|                 C|\n",
      "|   14214956681|               EWR|               EWR|       F|02/24/2017 10:00:00|02/26/2017 10:00:00|              NP|02/16/2017 01:03:02|          ZI|           2|       E|         USD|            GD|            RB|5400.0|       10800.0|       15524.0|                O|             UNL|02/16/2017 00:32:46|02/24/2017 10:00:00|                    US|           A|                 C|\n",
      "+--------------+------------------+------------------+--------+-------------------+-------------------+----------------+-------------------+------------+------------+--------+------------+--------------+--------------+------+--------------+--------------+-----------------+----------------+-------------------+-------------------+----------------------+------------+------------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "pricingDF.show(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ** Doing Date/Time Analysis**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Notice that the date or time columns are currently being interpreted as strings, rather than date or time objects:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- UniqueIdentity: string (nullable = true)\n",
      " |-- PickupLocationCode: string (nullable = true)\n",
      " |-- ReturnLocationCode: string (nullable = true)\n",
      " |-- CarClass: string (nullable = true)\n",
      " |-- PickupDate: string (nullable = true)\n",
      " |-- ReturnDate: string (nullable = true)\n",
      " |-- PricingSegmentId: string (nullable = true)\n",
      " |-- ShopDate: string (nullable = true)\n",
      " |-- CompetitorId: string (nullable = true)\n",
      " |-- LengthOfRent: integer (nullable = true)\n",
      " |-- PlanCode: string (nullable = true)\n",
      " |-- CurrencyCode: string (nullable = true)\n",
      " |-- ShopSourceCode: string (nullable = true)\n",
      " |-- ShopVendorCode: string (nullable = true)\n",
      " |-- Rate: double (nullable = true)\n",
      " |-- TimeAndMileage: double (nullable = true)\n",
      " |-- EstimatedTotal: double (nullable = true)\n",
      " |-- LocationIndicator: string (nullable = true)\n",
      " |-- MileageAllowance: string (nullable = true)\n",
      " |-- CreatedDate: string (nullable = true)\n",
      " |-- EndPickupDate: string (nullable = true)\n",
      " |-- PointOfSaleCountryCode: string (nullable = true)\n",
      " |-- CategoryCode: string (nullable = true)\n",
      " |-- RealTimeCanvasCode: string (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "pricingDF.printSchema()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's use the unix_timestamp() function to convert the string into a timestamp:\n",
    "\n",
    "https://people.apache.org/~pwendell/spark-nightly/spark-master-docs/latest/api/python/pyspark.sql.html?highlight=spark#pyspark.sql.functions.from_unixtime"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from pyspark.sql.functions import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Note that PySpark uses the Java Simple Date Format patterns\n",
    "\n",
    "from_pattern = 'MM/dd/yyyy hh:mm:ss'\n",
    "\n",
    "pricingTsDF = pricingDF \\\n",
    "  .withColumn('PickupDateTS', unix_timestamp(pricingDF['PickupDate'], from_pattern).cast(\"timestamp\")) \\\n",
    "  .drop('PickupDate') \\\n",
    "  .withColumn('ReturnDateTS', unix_timestamp(pricingDF['ReturnDate'], from_pattern).cast(\"timestamp\")) \\\n",
    "  .drop('ReturnDate') \\\n",
    "  .withColumn('ShopDateTS', unix_timestamp(pricingDF['ShopDate'], from_pattern).cast(\"timestamp\")) \\\n",
    "  .drop('ShopDate') \\\n",
    "  .withColumn('CreatedDateTS', unix_timestamp(pricingDF['CreatedDate'], from_pattern).cast(\"timestamp\")) \\\n",
    "  .drop('CreatedDate') \\\n",
    "  .withColumn('EndPickupDateTS', unix_timestamp(pricingDF['EndPickupDate'], from_pattern).cast(\"timestamp\")) \\\n",
    "  .drop('EndPickupDate') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- UniqueIdentity: string (nullable = true)\n",
      " |-- PickupLocationCode: string (nullable = true)\n",
      " |-- ReturnLocationCode: string (nullable = true)\n",
      " |-- CarClass: string (nullable = true)\n",
      " |-- PricingSegmentId: string (nullable = true)\n",
      " |-- CompetitorId: string (nullable = true)\n",
      " |-- LengthOfRent: integer (nullable = true)\n",
      " |-- PlanCode: string (nullable = true)\n",
      " |-- CurrencyCode: string (nullable = true)\n",
      " |-- ShopSourceCode: string (nullable = true)\n",
      " |-- ShopVendorCode: string (nullable = true)\n",
      " |-- Rate: double (nullable = true)\n",
      " |-- TimeAndMileage: double (nullable = true)\n",
      " |-- EstimatedTotal: double (nullable = true)\n",
      " |-- LocationIndicator: string (nullable = true)\n",
      " |-- MileageAllowance: string (nullable = true)\n",
      " |-- PointOfSaleCountryCode: string (nullable = true)\n",
      " |-- CategoryCode: string (nullable = true)\n",
      " |-- RealTimeCanvasCode: string (nullable = true)\n",
      " |-- PickupDateTS: timestamp (nullable = true)\n",
      " |-- ReturnDateTS: timestamp (nullable = true)\n",
      " |-- ShopDateTS: timestamp (nullable = true)\n",
      " |-- CreatedDateTS: timestamp (nullable = true)\n",
      " |-- EndPickupDateTS: timestamp (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# new date type columns are moved to the end of the column order\n",
    "pricingTsDF.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+--------------------+\n",
      "|month(CreatedDateTS)|\n",
      "+--------------------+\n",
      "|                   2|\n",
      "|                   3|\n",
      "+--------------------+\n",
      "\n",
      "CPU times: user 12 ms, sys: 4 ms, total: 16 ms\n",
      "Wall time: 41.6 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "#now we can do things like this\n",
    "pricingTsDF.select(month('CreatedDateTS')).distinct().orderBy('month(CreatedDateTS)').show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ** Memory, Caching and write to Parquet**\n",
    "\n",
    "Let's go to the SparkUI here: http://13.66.31.97:4040 (use your VM public IP) and understand what's happening inside the guts of Spark"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In real use case scenario, an executor can hold multiple number of partitions depending on the memory configured for each executor. Generally a cluster of ‘n’ executors with each executor having ‘c’ computational threads (cores) should have n*c partitions. However  it should be noted that above way of calculating number of partitions may not fit well for all size of input data for a given size of cluster. Another way of getting optimal partitions count is by keeping each partition size between 50MB to 200MB. It is advised that a task should not take more than 5 seconds to complete. If average time of task completion exceeds more than 5 seconds then RDD should be repartitioned to decrease partitions size. It should also be kept in mind that RDD repartition will cause shuffle and hence will increase execution time.\n",
    "\n",
    "By default 60% of memory allocated to spark application is used for RDD caching, 20% is used of data shuffle and rest 20% is used for storing objects created for task execution. It is highly recommended not to configure executor with more than 40G of  memory as higher memory allocation for executor can lead to large Garbage Collection (GC) pauses. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "36"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pricingTsDF.rdd.getNumPartitions()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pricingTsDF.createOrReplaceTempView(\"pricingVIEW\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "spark.catalog.cacheTable(\"pricingVIEW\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 12 ms, sys: 8 ms, total: 20 ms\n",
      "Wall time: 1min 38s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "25354815"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "# Call .count() to materialize the cache\n",
    "spark.table(\"pricingVIEW\").count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "25354815"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Note that the full scan + count in memory takes < 1 second!\n",
    "pricingTsDF.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----------+-----------+\n",
      "|  tableName|isTemporary|\n",
      "+-----------+-----------+\n",
      "|pricingview|       true|\n",
      "+-----------+-----------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "spark.sql(\"SHOW TABLES\").show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's see if we can improve times with repartitioning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "pricingTsDF.repartition(18).createOrReplaceTempView(\"pricingVIEW2\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 8 ms, sys: 0 ns, total: 8 ms\n",
      "Wall time: 36.7 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "spark.catalog.cacheTable(\"pricingVIEW2\")\n",
    "spark.table(\"pricingVIEW2\").count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pricingTsDF2 = spark.table(\"pricingVIEW2\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Look at the partitions size in the STORAGE section of the Spark UI. They look now closer to 100MB each. This is a best practice.\n",
    "\n",
    "Let's save now this partitions to a new parquet file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "azureml\r\n",
      "caffe2\r\n",
      "CNTK\r\n",
      "derby.log\r\n",
      "DocumentDBSample.ipynb\r\n",
      "h2o\r\n",
      "IDEAR.ipynb\r\n",
      "Introduction to Azure ML R notebooks.ipynb\r\n",
      "Introduction to Microsoft R Operationalization.ipynb\r\n",
      "IntroToJupyterPython.ipynb\r\n",
      "IntroTutorialinMicrosoftR.ipynb\r\n",
      "IntroTutorialinR.ipynb\r\n",
      "IrisClassifierPyMLWebService.ipynb\r\n",
      "julia\r\n",
      "LoadDataIntoDW.ipynb\r\n",
      "Meetup4-MDBDS-DSVM.ipynb\r\n",
      "metastore_db\r\n",
      "mxnet\r\n",
      "Pricing\r\n",
      "SparkML\r\n",
      "SQLDW_Explorations.ipynb\r\n",
      "tensorflow\r\n"
     ]
    }
   ],
   "source": [
    "!ls"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4 ms, sys: 0 ns, total: 4 ms\n",
      "Wall time: 25.7 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pricingTsDF2.write.format('parquet').mode('overwrite').save(\"./Parquet/COMPETITOR_RATE_20170216.parquet\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now the directory should contain 18 .gz compressed Parquet files (one for each partition):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total 555468\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31579869 Apr 21 19:19 part-r-00000-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31594877 Apr 21 19:19 part-r-00001-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31592691 Apr 21 19:19 part-r-00002-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31623386 Apr 21 19:19 part-r-00003-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31622641 Apr 21 19:19 part-r-00004-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31591826 Apr 21 19:19 part-r-00005-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31599357 Apr 21 19:19 part-r-00006-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31588931 Apr 21 19:19 part-r-00007-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31584509 Apr 21 19:19 part-r-00008-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31595548 Apr 21 19:19 part-r-00009-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31593927 Apr 21 19:19 part-r-00010-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31619524 Apr 21 19:19 part-r-00011-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31578189 Apr 21 19:19 part-r-00012-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31605998 Apr 21 19:19 part-r-00013-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31583784 Apr 21 19:19 part-r-00014-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31608153 Apr 21 19:19 part-r-00015-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31600099 Apr 21 19:19 part-r-00016-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin 31602248 Apr 21 19:19 part-r-00017-006261e2-400b-4743-bb26-dc3f14aca604.snappy.parquet\r\n",
      "-rw-r--r-- 1 pablomarin pablomarin        0 Apr 21 19:19 _SUCCESS\r\n"
     ]
    }
   ],
   "source": [
    "!ls -l ./Parquet/COMPETITOR_RATE_20170216.parquet"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here's how you can easily read the parquet file from Azure Storage in the future:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "tempDF = spark.read.parquet('./Parquet/COMPETITOR_RATE_20170216.parquet')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 0 ns, sys: 0 ns, total: 0 ns\n",
      "Wall time: 229 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "25354815"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "tempDF.count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **Analysis with PySpark DataFrames API**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### ![Spark Operations](http://curriculum-release.s3-website-us-west-2.amazonaws.com/wiki-book/general/spark_ta.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "DataFrames support two types of operations: *transformations* and *actions*.\n",
    "\n",
    "Transformations, like `select()` or `filter()` create a new DataFrame from an existing one.\n",
    "\n",
    "Actions, like `show()` or `count()`, return a value with results to the user. Other actions like `save()` write the DataFrame to distributed storage (like S3 or HDFS)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### ![Spark T/A](http://curriculum-release.s3-website-us-west-2.amazonaws.com/wiki-book/pagecounts/trans_and_actions.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Transformations contribute to a query plan,  but  nothing is executed until an action is called."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Let's see how many nulls there are"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 32 ms, sys: 12 ms, total: 44 ms\n",
      "Wall time: 4.73 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[u'{\"UniqueIdentity\":0,\"PickupLocationCode\":0,\"ReturnLocationCode\":0,\"CarClass\":0,\"PricingSegmentId\":0,\"CompetitorId\":0,\"LengthOfRent\":0,\"PlanCode\":0,\"CurrencyCode\":61,\"ShopSourceCode\":0,\"ShopVendorCode\":0,\"Rate\":0,\"TimeAndMileage\":0,\"EstimatedTotal\":0,\"LocationIndicator\":18965186,\"MileageAllowance\":61,\"PointOfSaleCountryCode\":0,\"CategoryCode\":0,\"RealTimeCanvasCode\":0,\"PickupDateTS\":0,\"ReturnDateTS\":0,\"ShopDateTS\":0,\"CreatedDateTS\":0,\"EndPickupDateTS\":0}']"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "pricingTsDF2.agg(*(sum(col(c).isNull().cast(\"integer\")).alias(c) for c in pricingTsDF2.columns)).toJSON().take(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets start by checking how many unique values each column has"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 64 ms, sys: 12 ms, total: 76 ms\n",
      "Wall time: 5min 37s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[u'{\"UniqueIdentity\":25354815,\"PickupLocationCode\":1199,\"ReturnLocationCode\":1247,\"CarClass\":19,\"PricingSegmentId\":7,\"CompetitorId\":36,\"LengthOfRent\":63,\"PlanCode\":4,\"CurrencyCode\":6,\"ShopSourceCode\":19,\"ShopVendorCode\":4,\"Rate\":115820,\"TimeAndMileage\":194133,\"EstimatedTotal\":234170,\"LocationIndicator\":1,\"MileageAllowance\":47,\"PointOfSaleCountryCode\":17,\"CategoryCode\":13,\"RealTimeCanvasCode\":3,\"PickupDateTS\":3694,\"ReturnDateTS\":4077,\"ShopDateTS\":49785,\"CreatedDateTS\":25688,\"EndPickupDateTS\":3709}']"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "pricingTsDF2.agg(*(countDistinct(col(c)).alias(c) for c in pricingTsDF2.columns)).toJSON().take(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's see how many bookings there are per country"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+----------------------+--------+\n",
      "|PointOfSaleCountryCode|   count|\n",
      "+----------------------+--------+\n",
      "|                    US|23663084|\n",
      "|                    CA|  759446|\n",
      "|                    DE|   81070|\n",
      "|                    GB|   79553|\n",
      "|                    CH|   78891|\n",
      "|                    FR|   77572|\n",
      "|                    IT|   77463|\n",
      "|                    UY|   77149|\n",
      "|                    NL|   77117|\n",
      "|                    SA|   74784|\n",
      "|                    AR|   72125|\n",
      "|                    AU|   59407|\n",
      "|                    NO|   48703|\n",
      "|                    AT|   36921|\n",
      "|                    SG|   32145|\n",
      "|                    CN|   29900|\n",
      "|                    JP|   29485|\n",
      "+----------------------+--------+\n",
      "\n",
      "CPU times: user 4 ms, sys: 4 ms, total: 8 ms\n",
      "Wall time: 1.03 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pricingTsDF2.select('PointOfSaleCountryCode').groupBy('PointOfSaleCountryCode').count().orderBy(\"count\", ascending=False).show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's see how many bookings are made where Pickup is different than return location"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------------+----------------------+------+\n",
      "|PickupLocationCode|PointOfSaleCountryCode| count|\n",
      "+------------------+----------------------+------+\n",
      "|               LAX|                    US|159999|\n",
      "|               BOS|                    US|123375|\n",
      "|               ATL|                    US|121420|\n",
      "|               MCO|                    US|106321|\n",
      "|               DCA|                    US|100808|\n",
      "|               SFO|                    US| 90005|\n",
      "|               IAD|                    US| 81717|\n",
      "|               LAS|                    US| 77067|\n",
      "|               BWI|                    US| 75265|\n",
      "|               TPA|                    US| 73799|\n",
      "|               MIA|                    US| 71903|\n",
      "|               PHL|                    US| 70824|\n",
      "|               DEN|                    US| 70087|\n",
      "|               ORD|                    US| 68302|\n",
      "|               EWR|                    US| 67989|\n",
      "|               SAN|                    US| 65464|\n",
      "|               SNA|                    US| 64792|\n",
      "|               JFK|                    US| 61372|\n",
      "|               RSW|                    US| 60055|\n",
      "|               FLL|                    US| 58722|\n",
      "+------------------+----------------------+------+\n",
      "only showing top 20 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "pricingTsDF2.filter(pricingTsDF2.PickupLocationCode <> pricingTsDF2.ReturnLocationCode).groupBy(\"PickupLocationCode\", \"PointOfSaleCountryCode\").count().orderBy(\"count\", ascending=False).show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## **SQL Queries and Plotting**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's explore the data a bit more, but this time using SQL."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "from datetime import datetime\n",
    "from matplotlib.pyplot import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's plot the rate changes during a day for L car class in the major city Los Angeles (LAX) where pickup = return and the lenght of rental was 3 days and pricing segment is \"ROD\" with Category \"S\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "pandas_df = spark.sql(\"\"\"\n",
    "SELECT CompetitorId, CarClass, Rate/100 as Rate, LengthOfRent, PricingSegmentId, CreatedDateTS \\\n",
    "            FROM pricingview2 \\\n",
    "            WHERE PickupLocationCode ==  ReturnLocationCode and PickupLocationCode == 'LAX' \\\n",
    "            and CarClass in ('L') \\\n",
    "            and LengthOfRent == 3 \\\n",
    "            and PricingSegmentId in ('ROD') and CategoryCode == 'S' \\\n",
    "            SORT BY CreatedDateTS\n",
    "        \"\"\").toPandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAg4AAAFjCAYAAABVOYMgAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkRozHAzDMAzDSM3/AgNDMIt3ZjPFAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2254ae0290>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "plot_date(pandas_df.CreatedDateTS, pandas_df.Rate)\n",
    "plt.gcf().autofmt_xdate()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's see the average price of each Competitor during this day, for rentals in LAX, 3 days, M Class, ROD Price Segment and S Category Code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "pandas_df = spark.sql(\"\"\"\n",
    "SELECT CompetitorId, avg(Rate)/100 as Rate\\\n",
    "            FROM pricingview2 \\\n",
    "            WHERE PickupLocationCode ==  ReturnLocationCode and PickupLocationCode == 'LAX' \\\n",
    "            and CarClass in ('M') \\\n",
    "            and LengthOfRent == 3 \\\n",
    "            and PricingSegmentId in ('ROD') and CategoryCode == 'S' \\\n",
    "            group by CompetitorId\n",
    "        \"\"\").toPandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhoAAAGTCAYAAABu7rurAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f22444ccf50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "# HISTOGRAM OF TIP AMOUNTS AND PASSENGER COUNT\n",
    "ax1 = pandas_df[['Rate']].plot(kind='bar')\n",
    "ax1.set_xticklabels(pandas_df.CompetitorId)\n",
    "ax1.set_title('Avg Rate per Competitor')\n",
    "ax1.set_xlabel('Competitor ID')\n",
    "ax1.set_ylabel('Rate ($)')\n",
    "plt.suptitle('')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## You made it to the end of the lab Part 1.\n",
    "## Next lab we will take this data set and start predicting "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2 Spark - local",
   "language": "python",
   "name": "spark-python"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}