Week_5_Song_recommender.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#Building a song recommender\n",
    "\n",
    "\n",
    "#Fire up GraphLab Create"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import graphlab"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#Load music data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "song_data = graphlab.SFrame('song_data.gl/')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#Explore data\n",
    "\n",
    "Music data shows how many times a user listened to a song, as well as the details of the song."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "song_data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##Showing the most popular songs in the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "graphlab.canvas.set_target('ipynb')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "song_data['song'].show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "len(song_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##Count number of unique users in the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "users = song_data['user_id'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "len(users)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#Create a song recommender"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_data,test_data = song_data.random_split(.8,seed=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##Simple popularity-based recommender"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "popularity_model = graphlab.popularity_recommender.create(train_data,\n",
    "                                                         user_id='user_id',\n",
    "                                                         item_id='song')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###Use the popularity model to make some predictions\n",
    "\n",
    "A popularity model makes the same prediction for all users, so provides no personalization."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "popularity_model.recommend(users=[users[0]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "popularity_model.recommend(users=[users[1]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##Build a song recommender with personalization\n",
    "\n",
    "We now create a model that allows us to make personalized recommendations to each user. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "personalized_model = graphlab.item_similarity_recommender.create(train_data,\n",
    "                                                                user_id='user_id',\n",
    "                                                                item_id='song')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###Applying the personalized model to make song recommendations\n",
    "\n",
    "As you can see, different users get different recommendations now."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "personalized_model.recommend(users=[users[0]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "personalized_model.recommend(users=[users[1]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###We can also apply the model to find similar songs to any song in the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "personalized_model.get_similar_items(['With Or Without You - U2'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "personalized_model.get_similar_items(['Chan Chan (Live) - Buena Vista Social Club'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#Quantitative comparison between the models\n",
    "\n",
    "We now formally compare the popularity and the personalized models using precision-recall curves. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "if graphlab.version[:3] >= \"1.6\":\n",
    "    model_performance = graphlab.compare(test_data, [popularity_model, personalized_model], user_sample=0.05)\n",
    "    graphlab.show_comparison(model_performance,[popularity_model, personalized_model])\n",
    "else:\n",
    "    %matplotlib inline\n",
    "    model_performance = graphlab.recommender.util.compare_models(test_data, [popularity_model, personalized_model], user_sample=.05)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "The curve shows that the personalized model provides much better performance. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}