evaluation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# TAAR – Evaluating existing recommenders\n",
    "\n",
    "Not every recommender can always make a recommendation. To evaluate the individual recommenders for the ensemble, we want to find out how often this is the case and how well the recommenders complement each other.\n",
    "\n",
    "This notebook either needs to be executed in the [TAAR](http://github.com/mozilla/taar) repository or somewhere where TAAR is in the Python path, because some TAAR recommenders are loaded in."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Retrieving the relevant variables from the longitudinal dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 12 ms, sys: 0 ns, total: 12 ms\n",
      "Wall time: 1min 14s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "frame = sqlContext.sql(\"\"\"\n",
    "WITH valid_clients AS (\n",
    "    SELECT *\n",
    "    FROM longitudinal\n",
    "    WHERE normalized_channel='release' AND build IS NOT NULL AND build[0].application_name='Firefox'\n",
    "),\n",
    "\n",
    "addons AS (\n",
    "    SELECT client_id, feature_row.*\n",
    "    FROM valid_clients\n",
    "    LATERAL VIEW explode(active_addons[0]) feature_row\n",
    "),\n",
    "    \n",
    "non_system_addons AS(\n",
    "    SELECT client_id, collect_set(key) AS installed_addons\n",
    "    FROM addons\n",
    "    WHERE NOT value.is_system\n",
    "    GROUP BY client_id\n",
    ")\n",
    "\n",
    "SELECT\n",
    "    l.client_id,\n",
    "    non_system_addons.installed_addons,\n",
    "    settings[0].locale AS locale,\n",
    "    geo_city[0] AS geoCity,\n",
    "    subsession_length[0] AS subsessionLength,\n",
    "    system_os[0].name AS os,\n",
    "    scalar_parent_browser_engagement_total_uri_count[0].value AS total_uri,\n",
    "    scalar_parent_browser_engagement_tab_open_event_count[0].value as tab_open_count,\n",
    "    places_bookmarks_count[0].sum as bookmark_count,\n",
    "    scalar_parent_browser_engagement_unique_domains_count[0].value as unique_tlds,\n",
    "    profile_creation_date[0] as profile_date,\n",
    "    submission_date[0] as submission_date\n",
    "FROM valid_clients l LEFT OUTER JOIN non_system_addons\n",
    "ON l.client_id = non_system_addons.client_id\n",
    "\"\"\")\n",
    "\n",
    "rdd = frame.rdd.cache()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Loading addon data (AMO)\n",
    "\n",
    "We need to load the addon database to find out which addons are considered useful by TAAR."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import boto3\n",
    "import json\n",
    "import logging\n",
    "\n",
    "from botocore.exceptions import ClientError\n",
    "\n",
    "logging.basicConfig(level=logging.INFO)\n",
    "logger = logging.getLogger(__name__)\n",
    "\n",
    "AMO_DUMP_BUCKET = 'telemetry-parquet'\n",
    "AMO_DUMP_KEY = 'telemetry-ml/addon_recommender/addons_database.json'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def load_amo_external_whitelist():\n",
    "    \"\"\" Download and parse the AMO add-on whitelist.\n",
    "    :raises RuntimeError: the AMO whitelist file cannot be downloaded or contains\n",
    "                          no valid add-ons.\n",
    "    \"\"\"\n",
    "    final_whitelist = []\n",
    "    amo_dump = {}\n",
    "    try:\n",
    "        # Load the most current AMO dump JSON resource.\n",
    "        s3 = boto3.client('s3')\n",
    "        s3_contents = s3.get_object(Bucket=AMO_DUMP_BUCKET, Key=AMO_DUMP_KEY)\n",
    "        amo_dump = json.loads(s3_contents['Body'].read())\n",
    "    except ClientError:\n",
    "        logger.exception(\"Failed to download from S3\", extra={\n",
    "            \"bucket\": AMO_DUMP_BUCKET,\n",
    "            \"key\": AMO_DUMP_KEY})\n",
    "\n",
    "    # If the load fails, we will have an empty whitelist, this may be problematic.\n",
    "    for key, value in amo_dump.items():\n",
    "        addon_files = value.get('current_version', {}).get('files', {})\n",
    "        # If any of the addon files are web_extensions compatible, it can be recommended.\n",
    "        if any([f.get(\"is_webextension\", False) for f in addon_files]):\n",
    "            final_whitelist.append(value['guid'])\n",
    "\n",
    "    if len(final_whitelist) == 0:\n",
    "        raise RuntimeError(\"Empty AMO whitelist detected\")\n",
    "\n",
    "    return final_whitelist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:botocore.vendored.requests.packages.urllib3.connectionpool:Starting new HTTP connection (1): 169.254.169.254\n",
      "INFO:botocore.vendored.requests.packages.urllib3.connectionpool:Starting new HTTP connection (1): 169.254.169.254\n",
      "INFO:botocore.vendored.requests.packages.urllib3.connectionpool:Starting new HTTPS connection (1): s3-us-west-2.amazonaws.com\n"
     ]
    }
   ],
   "source": [
    "whitelist = set(load_amo_external_whitelist())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Filtering out legacy addons \n",
    "\n",
    "This is a helper function that takes a list of addon IDs and only returns the IDs of addons that are useful for TAAR."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def get_whitelisted_addons(installed_addons):\n",
    "    return whitelist.intersection(installed_addons)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Completing client data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from dateutil.parser import parse as parse_date\n",
    "from datetime import datetime"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def compute_weeks_ago(formatted_date):\n",
    "    try:\n",
    "        date = parse_date(formatted_date).replace(tzinfo=None)\n",
    "    except ValueError: # raised when the date is in an unknown format\n",
    "        return float(\"inf\")\n",
    "    \n",
    "    days_ago = (datetime.today() - date).days\n",
    "    return days_ago / 7"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def complete_client_data(client_data):\n",
    "    client = client_data.asDict()\n",
    "    \n",
    "    addons = client['installed_addons'] or []\n",
    "    client['installed_addons'] = get_whitelisted_addons(addons)\n",
    "    client['disabled_addons_ids'] = addons\n",
    "    \n",
    "    client['locale'] = str(client['locale'])\n",
    "    client['profile_age_in_weeks'] = compute_weeks_ago(client['profile_date'])\n",
    "    client['submission_age_in_weeks'] = compute_weeks_ago(client['submission_date'])\n",
    "    \n",
    "    return client"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluating the existing recommenders\n",
    "\n",
    "To check if a recommender is able to make a recommendation, it's sometimes easier and cleaner to directly query it instead of checking the important attributes ourselves. For example, this is the case for the locale recommender."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sc.addPyFile(\"./taar/dist/mozilla_taar-0.0.16.dev15+g824aa58.d20171018-py2.7.egg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from taar.recommenders import CollaborativeRecommender, LegacyRecommender, LocaleRecommender"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class DummySimilarityRecommender:\n",
    "    def can_recommend(self, client_data):\n",
    "        REQUIRED_FIELDS = [\"geoCity\", \"subsessionLength\", \"locale\", \"os\", \"bookmark_count\", \"tab_open_count\",\n",
    "                           \"total_uri\", \"unique_tlds\"]\n",
    "\n",
    "        has_fields = all([client_data.get(f, None) is not None for f in REQUIRED_FIELDS])\n",
    "        return has_fields"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:requests.packages.urllib3.connectionpool:Starting new HTTPS connection (1): s3-us-west-2.amazonaws.com\n",
      "INFO:requests.packages.urllib3.connectionpool:Starting new HTTPS connection (1): s3-us-west-2.amazonaws.com\n",
      "INFO:boto3.resources.action:Calling s3:get_object with {u'Bucket': 'telemetry-parquet', u'Key': 'taar/legacy/legacy_dict.json'}\n",
      "INFO:botocore.vendored.requests.packages.urllib3.connectionpool:Starting new HTTPS connection (1): s3-us-west-2.amazonaws.com\n",
      "INFO:boto3.resources.action:Calling s3:get_object with {u'Bucket': 'telemetry-parquet', u'Key': 'taar/locale/top10_dict.json'}\n",
      "INFO:botocore.vendored.requests.packages.urllib3.connectionpool:Starting new HTTPS connection (1): s3-us-west-2.amazonaws.com\n"
     ]
    }
   ],
   "source": [
    "recommenders = {\n",
    "    \"collaborative\": CollaborativeRecommender(),\n",
    "    \"legacy\": LegacyRecommender(),\n",
    "    \"locale\": LocaleRecommender(\"./top_addons_by_locale.json\"),\n",
    "    \"similarity\": DummySimilarityRecommender()\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def test_recommenders(client):\n",
    "    return tuple([recommender.can_recommend(client) for recommender in recommenders.values()])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Computing combined counts"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We iterate over all clients in the longitudinal dataset, change the attributes to the expected format and then query the individual recommenders."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from operator import add\n",
    "from collections import defaultdict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "rdd_completed = rdd.map(complete_client_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def analyse(rdd):\n",
    "    results = rdd\\\n",
    "        .map(test_recommenders)\\\n",
    "        .map(lambda x: (x, 1))\\\n",
    "        .reduceByKey(add)\\\n",
    "        .collect()\n",
    "        \n",
    "    return defaultdict(int, results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1.3 s, sys: 116 ms, total: 1.41 s\n",
      "Wall time: 51.1 s\n"
     ]
    }
   ],
   "source": [
    "%time results = analyse(rdd_completed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "num_clients = sum(results.values())\n",
    "total_results = results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Computing individual counts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "individual_counts = []\n",
    "\n",
    "for i in range(len(recommenders)):\n",
    "    count = 0\n",
    "    \n",
    "    for key, key_count in results.items():\n",
    "        if key[i]:\n",
    "            count += key_count\n",
    "            \n",
    "    individual_counts.append(count)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Displaying the results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from pandas import DataFrame"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def format_int(num):\n",
    "    return \"{:,}\".format(num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def format_frequency(frequency):\n",
    "    return \"%.5f\" % frequency"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def get_relative_counts(counts, total=num_clients):\n",
    "    return [format_frequency(count / float(total)) for count in counts]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is a bit hacky. Sorting a data frame by formatted counts does not work; so we have to add the unformatted ones, sort the data frame, and then remove that column again."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def sorted_dataframe(df, order, key=\"unformatted_counts\"):\n",
    "    df[key] = order\n",
    "    return df.sort_values(by=key, ascending=False).drop(key, axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Individual counts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Relative count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>locale</th>\n",
       "      <td>0.95706</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>similarity</th>\n",
       "      <td>0.28744</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>collaborative</th>\n",
       "      <td>0.08311</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legacy</th>\n",
       "      <td>0.02739</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              Relative count\n",
       "locale               0.95706\n",
       "similarity           0.28744\n",
       "collaborative        0.08311\n",
       "legacy               0.02739"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = DataFrame(index=recommenders.keys(),\n",
    "          columns=[\"Relative count\"],\n",
    "          data=get_relative_counts(individual_counts)\n",
    ")\n",
    "\n",
    "sorted_dataframe(df, individual_counts)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$\\implies$ The locale and collaborative recommenders are able to generate recommendations most of the time. The legacy recommender can only make recommendations very seldomly as not many users seem to have (legacy) addons installed."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Combined counts\n",
    "\n",
    "It's interesting to see how well the individual recommenders complement each other. In the following, we count how often different combinations of the recommenders can make recommendations.\n",
    "\n",
    "The table is easier to read if cells are empty if a recommender is not available. If this is not desired, these variables can be changed:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "recommender_available_label = \"Available\"\n",
    "recommender_unavailable_label = \"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def format_labels(keys):\n",
    "    return tuple([recommender_available_label if key else recommender_unavailable_label for key in keys])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def format_data(keys, counts):\n",
    "    formatted_keys = map(format_labels, keys)\n",
    "    return [elems + count for elems, count in zip(formatted_keys, zip(*counts))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "columns = recommenders.keys() + [\"Relative counts\"]\n",
    "\n",
    "counts = get_relative_counts(results.values())\n",
    "data = format_data(results.keys(), [counts])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.62725</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.23407</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.03869</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.03081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.03016</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00894</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00603</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00553</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00082</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00027</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00023</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00019</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       locale     legacy collaborative similarity Relative counts\n",
       "2   Available                                             0.62725\n",
       "9   Available                           Available         0.23407\n",
       "12  Available                Available                    0.03869\n",
       "1   Available                Available  Available         0.03081\n",
       "3                                                         0.03016\n",
       "15                                      Available         0.00972\n",
       "4   Available  Available                                  0.00894\n",
       "0   Available  Available                Available         0.00603\n",
       "13  Available  Available     Available                    0.00575\n",
       "7   Available  Available     Available  Available         0.00553\n",
       "11                           Available                    0.00109\n",
       "14                           Available  Available         0.00082\n",
       "8              Available                                  0.00045\n",
       "6              Available                Available         0.00027\n",
       "10             Available     Available                    0.00023\n",
       "5              Available     Available  Available         0.00019"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = DataFrame(columns=columns, data=data)\n",
    "sorted_dataframe(df, results.values())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$\\implies$ If any recommender is available, then the locale recommenders is generally also available. Other than that, there is a good chance the the collaborative recommender is available.\n",
    "There is only a very small portion of cases where the similarity recommender was able to make a recommendation, when locale/collaborative were not; and not a single such case for the legacy recommender."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Grouped by number of available recommenders"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from itertools import groupby\n",
    "from operator import itemgetter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from IPython.display import display, Markdown"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "#### 0 available recommenders"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative to all</th>\n",
       "      <th>Relative to this table</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.03016</td>\n",
       "      <td>1.00000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  locale legacy collaborative similarity Relative to all  \\\n",
       "0                                                0.03016   \n",
       "\n",
       "  Relative to this table  \n",
       "0                1.00000  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### 1 available recommender"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative to all</th>\n",
       "      <th>Relative to this table</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.62725</td>\n",
       "      <td>0.98237</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00972</td>\n",
       "      <td>0.01522</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00109</td>\n",
       "      <td>0.00171</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00045</td>\n",
       "      <td>0.00070</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      locale     legacy collaborative similarity Relative to all  \\\n",
       "0  Available                                             0.62725   \n",
       "3                                      Available         0.00972   \n",
       "2                           Available                    0.00109   \n",
       "1             Available                                  0.00045   \n",
       "\n",
       "  Relative to this table  \n",
       "0                0.98237  \n",
       "3                0.01522  \n",
       "2                0.00171  \n",
       "1                0.00070  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### 2 available recommenders"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative to all</th>\n",
       "      <th>Relative to this table</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.23407</td>\n",
       "      <td>0.82705</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.03869</td>\n",
       "      <td>0.13671</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00894</td>\n",
       "      <td>0.03157</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00082</td>\n",
       "      <td>0.00290</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00027</td>\n",
       "      <td>0.00097</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00023</td>\n",
       "      <td>0.00081</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      locale     legacy collaborative similarity Relative to all  \\\n",
       "2  Available                           Available         0.23407   \n",
       "4  Available                Available                    0.03869   \n",
       "0  Available  Available                                  0.00894   \n",
       "5                           Available  Available         0.00082   \n",
       "1             Available                Available         0.00027   \n",
       "3             Available     Available                    0.00023   \n",
       "\n",
       "  Relative to this table  \n",
       "2                0.82705  \n",
       "4                0.13671  \n",
       "0                0.03157  \n",
       "5                0.00290  \n",
       "1                0.00097  \n",
       "3                0.00081  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### 3 available recommenders"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative to all</th>\n",
       "      <th>Relative to this table</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.03081</td>\n",
       "      <td>0.72010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00603</td>\n",
       "      <td>0.14099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00575</td>\n",
       "      <td>0.13443</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00019</td>\n",
       "      <td>0.00448</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      locale     legacy collaborative similarity Relative to all  \\\n",
       "1  Available                Available  Available         0.03081   \n",
       "0  Available  Available                Available         0.00603   \n",
       "3  Available  Available     Available                    0.00575   \n",
       "2             Available     Available  Available         0.00019   \n",
       "\n",
       "  Relative to this table  \n",
       "1                0.72010  \n",
       "0                0.14099  \n",
       "3                0.13443  \n",
       "2                0.00448  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### 4 available recommenders"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative to all</th>\n",
       "      <th>Relative to this table</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00553</td>\n",
       "      <td>1.00000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      locale     legacy collaborative similarity Relative to all  \\\n",
       "0  Available  Available     Available  Available         0.00553   \n",
       "\n",
       "  Relative to this table  \n",
       "0                1.00000  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for num, group in groupby(sorted(results.keys(), key=sum), sum):\n",
    "    display(Markdown(\"#### %d available recommender%s\" % (num, \"s\" if num != 1 else \"\")))\n",
    "    \n",
    "    sub_keys = list(group)\n",
    "    formatted_keys = map(format_labels, sub_keys)\n",
    "    \n",
    "    sub_counts = [results[key] for key in sub_keys]\n",
    "    sub_counts_to_total = get_relative_counts(sub_counts)\n",
    "    sub_counts_to_table = get_relative_counts(sub_counts, sum(sub_counts))\n",
    "    \n",
    "    zipped_data = zip(formatted_keys, sub_counts_to_total, sub_counts_to_table)\n",
    "    data = [elems + (counts, table_counts) for elems, counts, table_counts in zipped_data]\n",
    "    \n",
    "    columns = recommenders.keys() + [\"Relative to all\", \"Relative to this table\"]\n",
    "    \n",
    "    df = DataFrame(columns=columns, data=data)\n",
    "    df = sorted_dataframe(df, sub_counts)\n",
    "    display(df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## By dates\n",
    "\n",
    "In this section, we perform a similar analysis as before but on subsets of the data. These subsets are specified by when the client profile was generated. `conditions` is a list that contains ranges for the profile age in weeks. The end of the range is exclusive, similar to ranges in Python's standard library."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "conditions = [\n",
    "    (0, 1),\n",
    "    (1, 2),\n",
    "    (2, 3),\n",
    "    (3, 4)\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from numpy import argsort\n",
    "from itertools import product"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def attribute_between(attr, min_weeks, max_weeks):\n",
    "    return lambda client: min_weeks <= client[attr] < max_weeks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def get_conditioned_results(attr, conditions):\n",
    "    conditioned_results = {}\n",
    "\n",
    "    for (min_weeks, max_weeks) in conditions:\n",
    "        sub_rdd = rdd_completed.filter(attribute_between(attr, min_weeks, max_weeks))\n",
    "        conditioned_results[(min_weeks, max_weeks)] = analyse(sub_rdd)\n",
    "        \n",
    "    return conditioned_results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### By profile age in weeks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 5.86 s, sys: 388 ms, total: 6.25 s\n",
      "Wall time: 1min 3s\n"
     ]
    }
   ],
   "source": [
    "%time conditioned_results = get_conditioned_results(\"profile_age_in_weeks\", conditions)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To make things a little bit easier to read, only recommender combinations that actually appear are displayed in the table."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def nonzero_combinations(conditioned_results):\n",
    "    combinations = []\n",
    "\n",
    "    for sub_result in conditioned_results.values():\n",
    "        combinations += [key for key, value in sub_result.items() if value > 0]\n",
    "\n",
    "    return set(combinations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "combinations = nonzero_combinations(conditioned_results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "def display_individual_filtered_results(conditioned_results, combinations, label):\n",
    "    display(Markdown(\"### Filtering on the %s, Python-like exclusive ranges\" % label))\n",
    "\n",
    "    counts = []\n",
    "    titles = []\n",
    "\n",
    "    columns = recommenders.keys() + [\"Relative counts\"]\n",
    "\n",
    "    for key in conditions:\n",
    "        sub_results = conditioned_results[key]\n",
    "        values = [sub_results[sub_key] for sub_key in combinations]\n",
    "        summed = sum(values)\n",
    "\n",
    "        sub_counts = get_relative_counts(values, summed)\n",
    "        data = format_data(combinations, [sub_counts])\n",
    "        counts.append(sub_counts)\n",
    "\n",
    "        title = \"Between %d and %d weeks\" % key\n",
    "        titles.append(title)\n",
    "        display(Markdown(\"#### %s\" % title))\n",
    "\n",
    "        df = DataFrame(columns=columns, data=data)\n",
    "        df = sorted_dataframe(df, values)\n",
    "        display(df)\n",
    "\n",
    "    return counts, titles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "### Filtering on the profile age, Python-like exclusive ranges"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### Between 0 and 1 weeks"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.77052</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.15120</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.02525</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.02326</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01290</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00524</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00337</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00307</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00181</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00133</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00114</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00030</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00006</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       locale     legacy collaborative similarity Relative counts\n",
       "2   Available                                             0.77052\n",
       "10  Available                           Available         0.15120\n",
       "11                                                        0.02525\n",
       "3   Available                Available                    0.02326\n",
       "1   Available                Available  Available         0.01290\n",
       "15                                      Available         0.00524\n",
       "5   Available  Available                                  0.00337\n",
       "14  Available  Available     Available                    0.00307\n",
       "8   Available  Available     Available  Available         0.00181\n",
       "0   Available  Available                Available         0.00133\n",
       "12                           Available                    0.00114\n",
       "7                            Available  Available         0.00030\n",
       "6              Available     Available  Available         0.00024\n",
       "9              Available                                  0.00024\n",
       "4              Available     Available                    0.00006\n",
       "13             Available                Available         0.00006"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### Between 1 and 2 weeks"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.73305</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.18876</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.02495</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.01970</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01497</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00715</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00292</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00254</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00241</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00173</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00105</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00035</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00007</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       locale     legacy collaborative similarity Relative counts\n",
       "2   Available                                             0.73305\n",
       "10  Available                           Available         0.18876\n",
       "11                                                        0.02495\n",
       "3   Available                Available                    0.01970\n",
       "1   Available                Available  Available         0.01497\n",
       "15                                      Available         0.00715\n",
       "5   Available  Available                                  0.00292\n",
       "8   Available  Available     Available  Available         0.00254\n",
       "14  Available  Available     Available                    0.00241\n",
       "0   Available  Available                Available         0.00173\n",
       "12                           Available                    0.00105\n",
       "7                            Available  Available         0.00035\n",
       "9              Available                                  0.00015\n",
       "6              Available     Available  Available         0.00011\n",
       "13             Available                Available         0.00008\n",
       "4              Available     Available                    0.00007"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### Between 2 and 3 weeks"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.72011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.19623</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.02642</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.02061</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01699</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00717</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00295</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00291</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00283</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00212</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00058</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00043</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00016</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00014</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00011</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       locale     legacy collaborative similarity Relative counts\n",
       "2   Available                                             0.72011\n",
       "10  Available                           Available         0.19623\n",
       "11                                                        0.02642\n",
       "3   Available                Available                    0.02061\n",
       "1   Available                Available  Available         0.01699\n",
       "15                                      Available         0.00717\n",
       "5   Available  Available                                  0.00295\n",
       "8   Available  Available     Available  Available         0.00291\n",
       "14  Available  Available     Available                    0.00283\n",
       "0   Available  Available                Available         0.00212\n",
       "12                           Available                    0.00058\n",
       "7                            Available  Available         0.00043\n",
       "9              Available                                  0.00024\n",
       "4              Available     Available                    0.00016\n",
       "13             Available                Available         0.00014\n",
       "6              Available     Available  Available         0.00011"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### Between 3 and 4 weeks"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.70803</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.21223</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.02468</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.02100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01497</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00685</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00310</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00272</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00260</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00212</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00061</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00046</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00027</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00006</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       locale     legacy collaborative similarity Relative counts\n",
       "2   Available                                             0.70803\n",
       "10  Available                           Available         0.21223\n",
       "11                                                        0.02468\n",
       "3   Available                Available                    0.02100\n",
       "1   Available                Available  Available         0.01497\n",
       "15                                      Available         0.00685\n",
       "5   Available  Available                                  0.00310\n",
       "14  Available  Available     Available                    0.00272\n",
       "8   Available  Available     Available  Available         0.00260\n",
       "0   Available  Available                Available         0.00212\n",
       "12                           Available                    0.00061\n",
       "7                            Available  Available         0.00046\n",
       "9              Available                                  0.00027\n",
       "13             Available                Available         0.00024\n",
       "4              Available     Available                    0.00006\n",
       "6              Available     Available  Available         0.00006"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "counts, titles = display_individual_filtered_results(conditioned_results, combinations, label=\"profile age\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To make things a little bit easier to read, we can display all results in a single table."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def display_merged_filtered_results(counts, titles, total_results, combinations, label):\n",
    "    values = [total_results[sub_key] for sub_key in combinations]\n",
    "    sub_counts = get_relative_counts(values)\n",
    "    counts.append(sub_counts)\n",
    "    titles.append(\"Total, without any condition\")  \n",
    "\n",
    "    columns = recommenders.keys() + titles\n",
    "    data = format_data(combinations, counts)\n",
    "\n",
    "    df = DataFrame(columns=columns, data=data)\n",
    "    df = sorted_dataframe(df, counts[0])\n",
    "\n",
    "    display(Markdown(\"### Filtering on the %s, Python-like exclusive ranges – All in one table\" % label))\n",
    "    display(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "### Filtering on the profile age, Python-like exclusive ranges – All in one table"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Between 0 and 1 weeks</th>\n",
       "      <th>Between 1 and 2 weeks</th>\n",
       "      <th>Between 2 and 3 weeks</th>\n",
       "      <th>Between 3 and 4 weeks</th>\n",
       "      <th>Total, without any condition</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.77052</td>\n",
       "      <td>0.73305</td>\n",
       "      <td>0.72011</td>\n",
       "      <td>0.70803</td>\n",
       "      <td>0.62725</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.15120</td>\n",
       "      <td>0.18876</td>\n",
       "      <td>0.19623</td>\n",
       "      <td>0.21223</td>\n",
       "      <td>0.23407</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.02525</td>\n",
       "      <td>0.02495</td>\n",
       "      <td>0.02642</td>\n",
       "      <td>0.02468</td>\n",
       "      <td>0.03016</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.02326</td>\n",
       "      <td>0.01970</td>\n",
       "      <td>0.02061</td>\n",
       "      <td>0.02100</td>\n",
       "      <td>0.03869</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01290</td>\n",
       "      <td>0.01497</td>\n",
       "      <td>0.01699</td>\n",
       "      <td>0.01497</td>\n",
       "      <td>0.03081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00524</td>\n",
       "      <td>0.00715</td>\n",
       "      <td>0.00717</td>\n",
       "      <td>0.00685</td>\n",
       "      <td>0.00972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00337</td>\n",
       "      <td>0.00292</td>\n",
       "      <td>0.00295</td>\n",
       "      <td>0.00310</td>\n",
       "      <td>0.00894</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00307</td>\n",
       "      <td>0.00241</td>\n",
       "      <td>0.00283</td>\n",
       "      <td>0.00272</td>\n",
       "      <td>0.00575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00181</td>\n",
       "      <td>0.00254</td>\n",
       "      <td>0.00291</td>\n",
       "      <td>0.00260</td>\n",
       "      <td>0.00553</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00133</td>\n",
       "      <td>0.00173</td>\n",
       "      <td>0.00212</td>\n",
       "      <td>0.00212</td>\n",
       "      <td>0.00603</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00114</td>\n",
       "      <td>0.00105</td>\n",
       "      <td>0.00058</td>\n",
       "      <td>0.00061</td>\n",
       "      <td>0.00109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00030</td>\n",
       "      <td>0.00035</td>\n",
       "      <td>0.00043</td>\n",
       "      <td>0.00046</td>\n",
       "      <td>0.00082</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00024</td>\n",
       "      <td>0.00011</td>\n",
       "      <td>0.00011</td>\n",
       "      <td>0.00006</td>\n",
       "      <td>0.00019</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00024</td>\n",
       "      <td>0.00015</td>\n",
       "      <td>0.00024</td>\n",
       "      <td>0.00027</td>\n",
       "      <td>0.00045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00006</td>\n",
       "      <td>0.00007</td>\n",
       "      <td>0.00016</td>\n",
       "      <td>0.00006</td>\n",
       "      <td>0.00023</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00006</td>\n",
       "      <td>0.00008</td>\n",
       "      <td>0.00014</td>\n",
       "      <td>0.00024</td>\n",
       "      <td>0.00027</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       locale     legacy collaborative similarity Between 0 and 1 weeks  \\\n",
       "2   Available                                                   0.77052   \n",
       "10  Available                           Available               0.15120   \n",
       "11                                                              0.02525   \n",
       "3   Available                Available                          0.02326   \n",
       "1   Available                Available  Available               0.01290   \n",
       "15                                      Available               0.00524   \n",
       "5   Available  Available                                        0.00337   \n",
       "14  Available  Available     Available                          0.00307   \n",
       "8   Available  Available     Available  Available               0.00181   \n",
       "0   Available  Available                Available               0.00133   \n",
       "12                           Available                          0.00114   \n",
       "7                            Available  Available               0.00030   \n",
       "6              Available     Available  Available               0.00024   \n",
       "9              Available                                        0.00024   \n",
       "4              Available     Available                          0.00006   \n",
       "13             Available                Available               0.00006   \n",
       "\n",
       "   Between 1 and 2 weeks Between 2 and 3 weeks Between 3 and 4 weeks  \\\n",
       "2                0.73305               0.72011               0.70803   \n",
       "10               0.18876               0.19623               0.21223   \n",
       "11               0.02495               0.02642               0.02468   \n",
       "3                0.01970               0.02061               0.02100   \n",
       "1                0.01497               0.01699               0.01497   \n",
       "15               0.00715               0.00717               0.00685   \n",
       "5                0.00292               0.00295               0.00310   \n",
       "14               0.00241               0.00283               0.00272   \n",
       "8                0.00254               0.00291               0.00260   \n",
       "0                0.00173               0.00212               0.00212   \n",
       "12               0.00105               0.00058               0.00061   \n",
       "7                0.00035               0.00043               0.00046   \n",
       "6                0.00011               0.00011               0.00006   \n",
       "9                0.00015               0.00024               0.00027   \n",
       "4                0.00007               0.00016               0.00006   \n",
       "13               0.00008               0.00014               0.00024   \n",
       "\n",
       "   Total, without any condition  \n",
       "2                       0.62725  \n",
       "10                      0.23407  \n",
       "11                      0.03016  \n",
       "3                       0.03869  \n",
       "1                       0.03081  \n",
       "15                      0.00972  \n",
       "5                       0.00894  \n",
       "14                      0.00575  \n",
       "8                       0.00553  \n",
       "0                       0.00603  \n",
       "12                      0.00109  \n",
       "7                       0.00082  \n",
       "6                       0.00019  \n",
       "9                       0.00045  \n",
       "4                       0.00023  \n",
       "13                      0.00027  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display_merged_filtered_results(counts, titles, total_results, combinations, label=\"profile age\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### By submission date in weeks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 5.11 s, sys: 220 ms, total: 5.33 s\n",
      "Wall time: 1min 10s\n"
     ]
    }
   ],
   "source": [
    "%time conditioned_results_submission_date = get_conditioned_results(\"submission_age_in_weeks\", conditions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "### Filtering on the submission date, Python-like exclusive ranges"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### Between 0 and 1 weeks"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.42921</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.37171</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.06815</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.05122</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.01944</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01649</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01133</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01046</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00926</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00783</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00180</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00154</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00048</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00043</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00038</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00026</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       locale     legacy collaborative similarity Relative counts\n",
       "2   Available                                             0.42921\n",
       "10  Available                           Available         0.37171\n",
       "1   Available                Available  Available         0.06815\n",
       "3   Available                Available                    0.05122\n",
       "11                                                        0.01944\n",
       "15                                      Available         0.01649\n",
       "8   Available  Available     Available  Available         0.01133\n",
       "0   Available  Available                Available         0.01046\n",
       "5   Available  Available                                  0.00926\n",
       "14  Available  Available     Available                    0.00783\n",
       "7                            Available  Available         0.00180\n",
       "12                           Available                    0.00154\n",
       "13             Available                Available         0.00048\n",
       "9              Available                                  0.00043\n",
       "6              Available     Available  Available         0.00038\n",
       "4              Available     Available                    0.00026"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### Between 1 and 2 weeks"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.58081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.27560</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.04266</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.03264</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.02649</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01156</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00916</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00623</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00618</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00537</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00123</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00085</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00042</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00031</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00028</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00021</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       locale     legacy collaborative similarity Relative counts\n",
       "2   Available                                             0.58081\n",
       "10  Available                           Available         0.27560\n",
       "3   Available                Available                    0.04266\n",
       "1   Available                Available  Available         0.03264\n",
       "11                                                        0.02649\n",
       "15                                      Available         0.01156\n",
       "5   Available  Available                                  0.00916\n",
       "0   Available  Available                Available         0.00623\n",
       "14  Available  Available     Available                    0.00618\n",
       "8   Available  Available     Available  Available         0.00537\n",
       "12                           Available                    0.00123\n",
       "7                            Available  Available         0.00085\n",
       "9              Available                                  0.00042\n",
       "13             Available                Available         0.00031\n",
       "4              Available     Available                    0.00028\n",
       "6              Available     Available  Available         0.00021"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### Between 2 and 3 weeks"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.65590</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.21726</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.03932</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.02892</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.02249</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00951</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00862</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00584</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00520</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00414</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00113</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00059</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00018</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       locale     legacy collaborative similarity Relative counts\n",
       "2   Available                                             0.65590\n",
       "10  Available                           Available         0.21726\n",
       "3   Available                Available                    0.03932\n",
       "11                                                        0.02892\n",
       "1   Available                Available  Available         0.02249\n",
       "5   Available  Available                                  0.00951\n",
       "15                                      Available         0.00862\n",
       "14  Available  Available     Available                    0.00584\n",
       "0   Available  Available                Available         0.00520\n",
       "8   Available  Available     Available  Available         0.00414\n",
       "12                           Available                    0.00113\n",
       "7                            Available  Available         0.00059\n",
       "9              Available                                  0.00045\n",
       "13             Available                Available         0.00024\n",
       "4              Available     Available                    0.00020\n",
       "6              Available     Available  Available         0.00018"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "#### Between 3 and 4 weeks"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Relative counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.68416</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.19886</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.03622</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.02901</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01904</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00935</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00722</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00516</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00376</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00104</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00050</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00047</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00011</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       locale     legacy collaborative similarity Relative counts\n",
       "2   Available                                             0.68416\n",
       "10  Available                           Available         0.19886\n",
       "3   Available                Available                    0.03622\n",
       "11                                                        0.02901\n",
       "1   Available                Available  Available         0.01904\n",
       "5   Available  Available                                  0.00935\n",
       "15                                      Available         0.00722\n",
       "14  Available  Available     Available                    0.00516\n",
       "0   Available  Available                Available         0.00472\n",
       "8   Available  Available     Available  Available         0.00376\n",
       "12                           Available                    0.00104\n",
       "7                            Available  Available         0.00050\n",
       "9              Available                                  0.00047\n",
       "13             Available                Available         0.00020\n",
       "4              Available     Available                    0.00018\n",
       "6              Available     Available  Available         0.00011"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "### Filtering on the submission date, Python-like exclusive ranges – All in one table"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>locale</th>\n",
       "      <th>legacy</th>\n",
       "      <th>collaborative</th>\n",
       "      <th>similarity</th>\n",
       "      <th>Between 0 and 1 weeks</th>\n",
       "      <th>Between 1 and 2 weeks</th>\n",
       "      <th>Between 2 and 3 weeks</th>\n",
       "      <th>Between 3 and 4 weeks</th>\n",
       "      <th>Total, without any condition</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.42921</td>\n",
       "      <td>0.58081</td>\n",
       "      <td>0.65590</td>\n",
       "      <td>0.68416</td>\n",
       "      <td>0.62725</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.37171</td>\n",
       "      <td>0.27560</td>\n",
       "      <td>0.21726</td>\n",
       "      <td>0.19886</td>\n",
       "      <td>0.23407</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.06815</td>\n",
       "      <td>0.03264</td>\n",
       "      <td>0.02249</td>\n",
       "      <td>0.01904</td>\n",
       "      <td>0.03081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.05122</td>\n",
       "      <td>0.04266</td>\n",
       "      <td>0.03932</td>\n",
       "      <td>0.03622</td>\n",
       "      <td>0.03869</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.01944</td>\n",
       "      <td>0.02649</td>\n",
       "      <td>0.02892</td>\n",
       "      <td>0.02901</td>\n",
       "      <td>0.03016</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01649</td>\n",
       "      <td>0.01156</td>\n",
       "      <td>0.00862</td>\n",
       "      <td>0.00722</td>\n",
       "      <td>0.00972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01133</td>\n",
       "      <td>0.00537</td>\n",
       "      <td>0.00414</td>\n",
       "      <td>0.00376</td>\n",
       "      <td>0.00553</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.01046</td>\n",
       "      <td>0.00623</td>\n",
       "      <td>0.00520</td>\n",
       "      <td>0.00472</td>\n",
       "      <td>0.00603</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00926</td>\n",
       "      <td>0.00916</td>\n",
       "      <td>0.00951</td>\n",
       "      <td>0.00935</td>\n",
       "      <td>0.00894</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00783</td>\n",
       "      <td>0.00618</td>\n",
       "      <td>0.00584</td>\n",
       "      <td>0.00516</td>\n",
       "      <td>0.00575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00180</td>\n",
       "      <td>0.00085</td>\n",
       "      <td>0.00059</td>\n",
       "      <td>0.00050</td>\n",
       "      <td>0.00082</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00154</td>\n",
       "      <td>0.00123</td>\n",
       "      <td>0.00113</td>\n",
       "      <td>0.00104</td>\n",
       "      <td>0.00109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00048</td>\n",
       "      <td>0.00031</td>\n",
       "      <td>0.00024</td>\n",
       "      <td>0.00020</td>\n",
       "      <td>0.00027</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0.00043</td>\n",
       "      <td>0.00042</td>\n",
       "      <td>0.00045</td>\n",
       "      <td>0.00047</td>\n",
       "      <td>0.00045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td>0.00038</td>\n",
       "      <td>0.00021</td>\n",
       "      <td>0.00018</td>\n",
       "      <td>0.00011</td>\n",
       "      <td>0.00019</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td></td>\n",
       "      <td>Available</td>\n",
       "      <td>Available</td>\n",
       "      <td></td>\n",
       "      <td>0.00026</td>\n",
       "      <td>0.00028</td>\n",
       "      <td>0.00020</td>\n",
       "      <td>0.00018</td>\n",
       "      <td>0.00023</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       locale     legacy collaborative similarity Between 0 and 1 weeks  \\\n",
       "2   Available                                                   0.42921   \n",
       "10  Available                           Available               0.37171   \n",
       "1   Available                Available  Available               0.06815   \n",
       "3   Available                Available                          0.05122   \n",
       "11                                                              0.01944   \n",
       "15                                      Available               0.01649   \n",
       "8   Available  Available     Available  Available               0.01133   \n",
       "0   Available  Available                Available               0.01046   \n",
       "5   Available  Available                                        0.00926   \n",
       "14  Available  Available     Available                          0.00783   \n",
       "7                            Available  Available               0.00180   \n",
       "12                           Available                          0.00154   \n",
       "13             Available                Available               0.00048   \n",
       "9              Available                                        0.00043   \n",
       "6              Available     Available  Available               0.00038   \n",
       "4              Available     Available                          0.00026   \n",
       "\n",
       "   Between 1 and 2 weeks Between 2 and 3 weeks Between 3 and 4 weeks  \\\n",
       "2                0.58081               0.65590               0.68416   \n",
       "10               0.27560               0.21726               0.19886   \n",
       "1                0.03264               0.02249               0.01904   \n",
       "3                0.04266               0.03932               0.03622   \n",
       "11               0.02649               0.02892               0.02901   \n",
       "15               0.01156               0.00862               0.00722   \n",
       "8                0.00537               0.00414               0.00376   \n",
       "0                0.00623               0.00520               0.00472   \n",
       "5                0.00916               0.00951               0.00935   \n",
       "14               0.00618               0.00584               0.00516   \n",
       "7                0.00085               0.00059               0.00050   \n",
       "12               0.00123               0.00113               0.00104   \n",
       "13               0.00031               0.00024               0.00020   \n",
       "9                0.00042               0.00045               0.00047   \n",
       "6                0.00021               0.00018               0.00011   \n",
       "4                0.00028               0.00020               0.00018   \n",
       "\n",
       "   Total, without any condition  \n",
       "2                       0.62725  \n",
       "10                      0.23407  \n",
       "1                       0.03081  \n",
       "3                       0.03869  \n",
       "11                      0.03016  \n",
       "15                      0.00972  \n",
       "8                       0.00553  \n",
       "0                       0.00603  \n",
       "5                       0.00894  \n",
       "14                      0.00575  \n",
       "7                       0.00082  \n",
       "12                      0.00109  \n",
       "13                      0.00027  \n",
       "9                       0.00045  \n",
       "6                       0.00019  \n",
       "4                       0.00023  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "label = \"submission date\"\n",
    "combinations = nonzero_combinations(conditioned_results_submission_date)\n",
    "counts, titles = display_individual_filtered_results(conditioned_results_submission_date, combinations, label)\n",
    "display_merged_filtered_results(counts, titles, total_results, combinations, label)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Addon counts\n",
    "\n",
    "We want to train an ensemble model using the individual recommenders that we already have. To optimize this ensemble model, we need some training data. We have information about what addons different users have installed, so it would make sense to check if our ensemble model would also recommend these addons to the respective users.\n",
    "\n",
    "However, there is one fundamental conflict here: To be able to make recommendations, the collaborative recommender already needs some information about which addons a user has installed. Thus, we can only use a subset of a user's installed addons for evaluation. These addons are masked and the collaborative recommender then only uses the unmasked addons. The open question is how large this subset of masked addons should be.\n",
    "\n",
    "Choosing this size is a trade-off between three factors:\n",
    "1. By masking more addons, we give the evaluation function more data to work with\n",
    "2. If we mask fewer addons, the collaborative filter can make better recommendations\n",
    "3. There are not that many users who have many addons installed. This means we need to be careful not to make our evaluation set too biased. For example, if we always mask at least five addons, then only users with more than six addons could be part of the evaluation set, which is a small subset of the entire population\n",
    "\n",
    "To be able to make this decision, it's helpful to look at the distribution of the number of addons that users have installed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/mnt/anaconda2/lib/python2.7/site-packages/matplotlib/__init__.py:878: UserWarning: axes.color_cycle is deprecated and replaced with axes.prop_cycle; please use the latter.\n",
      "  warnings.warn(self.msg_depr % (key, alt_key))\n"
     ]
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "%matplotlib inline\n",
    "sns.set(style=\"darkgrid\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "addon_counts = rdd_completed\\\n",
    "    .map(lambda client: len(client['installed_addons']))\\\n",
    "    .map(lambda x: (x, 1))\\\n",
    "    .reduceByKey(add)\\\n",
    "    .collect()\n",
    "    \n",
    "addon_counts = sorted(addon_counts, key=itemgetter(0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def addon_counts_to_proportions(addon_counts):\n",
    "    num_addons, num_addons_count = zip(*addon_counts)\n",
    "    \n",
    "    num_addons_total = float(sum(num_addons_count))\n",
    "    num_addons_count = np.array(num_addons_count) / num_addons_total\n",
    "    \n",
    "    return num_addons, num_addons_count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def plot_addon_distribution(addon_counts):\n",
    "    num_addons, num_addons_count = addon_counts_to_proportions(addon_counts)\n",
    "    \n",
    "    plt.bar(num_addons, num_addons_count, width=1.)\n",
    "    plt.title(\"Number of addons per user\")\n",
    "    plt.xlabel(\"Number of addons\")\n",
    "    plt.ylabel(\"Proportion of users\")\n",
    "    plt.show()\n",
    "    \n",
    "    plt.plot(np.cumsum(num_addons_count))\n",
    "    plt.title(\"Number of addons per user (cumulative)\")\n",
    "    plt.xlabel(\"Number of addons\")\n",
    "    plt.ylabel(\"Proportion of users\")\n",
    "    plt.ylim(0, 1)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAr4AAAIACAYAAABkR29wAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe33ac21550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAr4AAAIACAYAAABkR29wAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe3480e8bd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_addon_distribution(addon_counts)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As we can see, there is a substantial number of users that have no addons installed. Nearly all other users have fewer than 10 addons installed. To make the plot a bit easier to read, it's helpful to hide the long tail."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAr4AAAIACAYAAABkR29wAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAjCGtWhYYpAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe3381b8b50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAr4AAAIACAYAAABkR29wAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nREREJAn/B8td8p8xFpZAAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe33bcbe410>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "reasonable_addon_counts = filter(lambda (num_addons, count): num_addons < 15, addon_counts)\n",
    "plot_addon_distribution(reasonable_addon_counts)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$\\implies$ To get an evaluation set of a decent size that's at least partly representative, it's important to include users that only have a few addons installed. One idea could be to mask half of the addons that users in the evaluation set have installed. This way, we would be able to include users that only have a few addons installed. If the number of addons is odd, we could uniformly randomly round up or down.\n",
    "\n",
    "The alternative would be to make the portion of masked addons a little bit higher or lower. This could make sense if we notice that the collaborative recommender is generally able to make much better recommendations using two addons than using one addon; or, the other way around, if we notice that it adds little value to evaluate based on a single addon."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Number of addons</th>\n",
       "      <th>Proportion</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0.91689</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.06219</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0.01440</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>0.00386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0.00142</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>0.00062</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>0.00028</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>0.00014</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>0.00008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>9</td>\n",
       "      <td>0.00004</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>10</td>\n",
       "      <td>0.00003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>11</td>\n",
       "      <td>0.00001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>12</td>\n",
       "      <td>0.00001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>13</td>\n",
       "      <td>0.00001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>14</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>15</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>16</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>17</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>18</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>19</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>20</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>21</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>22</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>23</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>27</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>28</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>31</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>36</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>45</td>\n",
       "      <td>0.00000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Number of addons Proportion\n",
       "0                  0    0.91689\n",
       "1                  1    0.06219\n",
       "2                  2    0.01440\n",
       "3                  3    0.00386\n",
       "4                  4    0.00142\n",
       "5                  5    0.00062\n",
       "6                  6    0.00028\n",
       "7                  7    0.00014\n",
       "8                  8    0.00008\n",
       "9                  9    0.00004\n",
       "10                10    0.00003\n",
       "11                11    0.00001\n",
       "12                12    0.00001\n",
       "13                13    0.00001\n",
       "14                14    0.00000\n",
       "15                15    0.00000\n",
       "16                16    0.00000\n",
       "17                17    0.00000\n",
       "18                18    0.00000\n",
       "19                19    0.00000\n",
       "20                20    0.00000\n",
       "21                21    0.00000\n",
       "22                22    0.00000\n",
       "23                23    0.00000\n",
       "24                27    0.00000\n",
       "25                28    0.00000\n",
       "26                31    0.00000\n",
       "27                36    0.00000\n",
       "28                45    0.00000"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_addons, num_addons_count = addon_counts_to_proportions(addon_counts)\n",
    "num_addons_count = map(format_frequency, num_addons_count)\n",
    "\n",
    "DataFrame(\n",
    "    columns=[\"Number of addons\", \"Proportion\"],\n",
    "    data=zip(num_addons, num_addons_count)\n",
    ").head(40)"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [default]",
   "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": "2.7.12"
  },
  "notify_time": "10",
  "toc": {
   "nav_menu": {},
   "number_sections": false,
   "sideBar": true,
   "skip_h1_title": false,
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": "block",
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}