{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Yoochoose: Aggregate Sessions\n",
    "\n",
    "Here we download the data, upload it to HDFS and create session objects using Spark."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Upload data to HDFS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--2015-07-07 15:18:40--  http://s3-eu-west-1.amazonaws.com/yc-rdata/yoochoose-data.7z\n",
      "Resolving s3-eu-west-1.amazonaws.com (s3-eu-west-1.amazonaws.com)... 54.231.136.80\n",
      "Connecting to s3-eu-west-1.amazonaws.com (s3-eu-west-1.amazonaws.com)|54.231.136.80|:80... connected.\n",
      "HTTP request sent, awaiting response... 200 OK\n",
      "Length: 287211932 (274M) [application/octet-stream]\n",
      "Saving to: `yoochoose-data.7z'\n",
      "\n",
      "100%[======================================>] 287,211,932 31.0M/s   in 14s     \n",
      "\n",
      "2015-07-07 15:18:54 (19.9 MB/s) - `yoochoose-data.7z' saved [287211932/287211932]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "!wget http://s3-eu-west-1.amazonaws.com/yc-rdata/yoochoose-data.7z"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "7-Zip [64] 9.20  Copyright (c) 1999-2010 Igor Pavlov  2010-11-18\n",
      "p7zip Version 9.20 (locale=en_US.UTF-8,Utf16=on,HugeFiles=on,8 CPUs)\n",
      "\n",
      "Processing archive: yoochoose-data.7z\n",
      "\n",
      "Extracting  yoochoose-buys.dat\n",
      "Extracting  yoochoose-clicks.dat\n",
      "Extracting  yoochoose-test.dat\n",
      "Extracting  dataset-README.txt\n",
      "\n",
      "Everything is Ok\n",
      "\n",
      "Files: 4\n",
      "Size:       1914111754\n",
      "Compressed: 287211932\n"
     ]
    }
   ],
   "source": [
    "!7z x yoochoose-data.7z"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Put data to HDFS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "!hdfs dfs -mkdir -p yoochoose/\n",
    "!hdfs dfs -put yoochoose-*.dat yoochoose/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "!rm yoochoose-* dataset-README.txt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Aggregate sessions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import datetime\n",
    "import operator\n",
    "\n",
    "def parse_datetime(dt_str):\n",
    "    return datetime.datetime.strptime(dt_str, '%Y-%m-%dT%H:%M:%S.%fZ')\n",
    "\n",
    "def parse_clicks(line):\n",
    "    parts = line.split(',')\n",
    "    session_id = int(parts[0])\n",
    "    timestamp, item_id, category = parse_datetime(parts[1]), parts[2], parts[3]\n",
    "    return session_id, (timestamp, item_id, category)\n",
    "\n",
    "def parse_buys(line):\n",
    "    parts = line.split(',')\n",
    "    session_id = int(parts[0])\n",
    "    timestamp, item_id, price, quantity = parse_datetime(parts[1]), parts[2], float(parts[3]), int(parts[4])\n",
    "    return session_id, (timestamp, item_id, price, quantity)\n",
    "\n",
    "def sort_sessions((session_id, (clicks, buys))):\n",
    "    clicks = sorted(clicks, key=operator.itemgetter(0)) if clicks is not None else []\n",
    "    buys = sorted(buys, key=operator.itemgetter(0)) if buys is not None else []\n",
    "    return session_id, (clicks, buys)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Run Spark job"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
     ]
    }
   ],
   "source": [
    "# Start Spark context\n",
    "%spark YoochooseSessions\n",
    "\n",
    "# read input\n",
    "clicks = sc.textFile('yoochoose/yoochoose-clicks.dat', 40).map(parse_clicks).groupByKey()\n",
    "clicksTest = sc.textFile('yoochoose/yoochoose-test.dat', 40).map(parse_clicks).groupByKey()\n",
    "buys = sc.textFile('yoochoose/yoochoose-buys.dat', 40).map(parse_buys).groupByKey()\n",
    "\n",
    "train_sessions = clicks.fullOuterJoin(buys).map(sort_sessions)\n",
    "test_sessions = clicksTest.map(lambda (session_id, clicks): (session_id, (clicks, None))).map(sort_sessions)\n",
    "\n",
    "# screate session files\n",
    "train_sessions.saveAsPickleFile('yoochoose/train_sessions.pickle')\n",
    "test_sessions.saveAsPickleFile('yoochoose/test_sessions.pickle')\n",
    "\n",
    "sc.stop()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}