HW4_solutions.ipynb

{
 "metadata": {
  "name": ""
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "HW4: Do we really need Chocolate Recommendations?"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "<img src=\"http://1.bp.blogspot.com/-8dGYKeMKNaU/TvutmCenc-I/AAAAAAAABEo/b2Czf4RlAzw/s1600/Death%2BBy%2BChocolate.JPG\" width=\"400\" height=\"300\"/>"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Before You Start\n",
      "\n",
      "This is a **long** homework. Please start early. It uses a lot of different (and sometimes complex) concepts, so you might find yourself reading a lot. So, please, give yourself a lot of time.\n",
      "\n",
      "Also, please see this [link](http://nbviewer.ipython.org/urls/raw.github.com/cs109/content/master/InstructionsForAmazonEMR.ipynb) on getting an Amazon Web Services account soon, so that you dont delay its creation. This class gives you $100 in credits which you will use for this homework, possibly your project, and any other projects you might like.\n",
      "\n",
      "Finally, please go to the labs. The one on 18th October (Today) will cover Gibbs Sampling and Bayesian Normal distributions. The one on the 25th will cover Map-Reduce. Both will help on the homework."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Collaborative Filtering systems\n",
      "\n",
      "In this homework, you will create a recommendation system for **restaurants** using [collaborative filtering](http://en.wikipedia.org/wiki/Collaborative_filtering) (CF). The general structure of a recommendation system is that there are users and there are items. Users express explicit or implicit preferences towards certain items. CF thus relies on users' past behavior.\n",
      "\n",
      "There are two primary approaches to CF: neighboorhood and latent factor model. The former is concerned with computing the relationships between items or between users. In the latter approach you have a model of hidden factors through which users and items are transformed to the same space. For example, if you are rating movies we may transform items into genre factors, and users into their preference for a particular genre.\n",
      "\n",
      "Factor models generally lead to more accurate recommenders. One of the reasons for this is the sparsity of the item-user matrix. Most users tend to rate barely one or two items. Latent factor models are more expressive, and fit fewer parameters. However, neighborhood models are more prevalent, as they have an intuitive aspect that appeals to users(if you liked this you will like that) and online(a new preference can be incorporated very quickly).\n",
      "\n",
      "Most recommenders today combine neighboorhood CF with model based CF, and SVD based matrix factorization approaches.\n",
      "\n",
      "To see the example of a simple beer recommender, go [here](http://nbviewer.ipython.org/20a18d52c539b87de2af). This homework is inspired by the one there but we go after food instead, and go deeper into the problem of making recommendations."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "### User and Item based approaches\n",
      "\n",
      "Original approaches to neighborhood based CF used user-user models. By this we mean that rating estimates are made from recorded ratings of like minded users. However, since most users tend to rate very few items, this is usually a losing proposition for explicit-rating based recommenders. Thus, most neighborhood based systems such as Amazon these days rely on item-item approaches. In these methods, a rating is estimated by other ratings made by the user on \"similar\" or \"nearby\" items: we have a K-Nearest-Neighbors algorithm, in effect."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Outline of this Homework\n",
      "\n",
      "The outline of this homework is as follows:\n",
      "\n",
      "1. Create a database of item-item similarities. Use this to implement a neighborhood-based CF recommender that can answer simple questions like \"give me more restaurants like this one\". This part of the homework assumes that the similaties calculated make good \"global recommendations\".\n",
      "\n",
      "2. In the second part, we go one step further and attempt to predict the rating that a user will give an item they have not seen before. This requires that we find the restaurants that *this* user would rate as similar (not just those which are globally similar). \n",
      "\n",
      "3. In the third part, we implement a factor-based CF recommender using a Bayesian model. While quite a bit more complex, this allows us to pool information both about similar users and about similar restaurants.\n",
      "\n",
      "5. We will scale up our system by creating a recommender on the lines of Q1 and Q2 that works on the entire data set. We will use the map-reduce paradigm to split the computation over multiple machines."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "You will start simply, by working on a subset of the restaurant data before generalizing to the entire data set in Problem 4. The complete data set has 150,000 reviews, but we shall start with just about 7000. You will create this  smaller set by taking all the users who had rated more than 60 restaurants, and all the businesses which had greater than 150 reviews from the larger data set. This is not a random set: indeed we use it as it a computationally tractable set that is a bit less sparse than the entire data set."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%matplotlib inline\n",
      "from collections import defaultdict\n",
      "import json\n",
      "\n",
      "import numpy as np\n",
      "import scipy as sp\n",
      "import matplotlib.pyplot as plt\n",
      "import pandas as pd\n",
      "\n",
      "from matplotlib import rcParams\n",
      "import matplotlib.cm as cm\n",
      "import matplotlib as mpl\n",
      "\n",
      "#colorbrewer2 Dark2 qualitative color table\n",
      "dark2_colors = [(0.10588235294117647, 0.6196078431372549, 0.4666666666666667),\n",
      "                (0.8509803921568627, 0.37254901960784315, 0.00784313725490196),\n",
      "                (0.4588235294117647, 0.4392156862745098, 0.7019607843137254),\n",
      "                (0.9058823529411765, 0.1607843137254902, 0.5411764705882353),\n",
      "                (0.4, 0.6509803921568628, 0.11764705882352941),\n",
      "                (0.9019607843137255, 0.6705882352941176, 0.00784313725490196),\n",
      "                (0.6509803921568628, 0.4627450980392157, 0.11372549019607843)]\n",
      "\n",
      "rcParams['figure.figsize'] = (10, 6)\n",
      "rcParams['figure.dpi'] = 150\n",
      "rcParams['axes.color_cycle'] = dark2_colors\n",
      "rcParams['lines.linewidth'] = 2\n",
      "rcParams['axes.facecolor'] = 'white'\n",
      "rcParams['font.size'] = 14\n",
      "rcParams['patch.edgecolor'] = 'white'\n",
      "rcParams['patch.facecolor'] = dark2_colors[0]\n",
      "rcParams['font.family'] = 'StixGeneral'\n",
      "\n",
      "\n",
      "def remove_border(axes=None, top=False, right=False, left=True, bottom=True):\n",
      "    \"\"\"\n",
      "    Minimize chartjunk by stripping out unnecesasry plot borders and axis ticks\n",
      "    \n",
      "    The top/right/left/bottom keywords toggle whether the corresponding plot border is drawn\n",
      "    \"\"\"\n",
      "    ax = axes or plt.gca()\n",
      "    ax.spines['top'].set_visible(top)\n",
      "    ax.spines['right'].set_visible(right)\n",
      "    ax.spines['left'].set_visible(left)\n",
      "    ax.spines['bottom'].set_visible(bottom)\n",
      "    \n",
      "    #turn off all ticks\n",
      "    ax.yaxis.set_ticks_position('none')\n",
      "    ax.xaxis.set_ticks_position('none')\n",
      "    \n",
      "    #now re-enable visibles\n",
      "    if top:\n",
      "        ax.xaxis.tick_top()\n",
      "    if bottom:\n",
      "        ax.xaxis.tick_bottom()\n",
      "    if left:\n",
      "        ax.yaxis.tick_left()\n",
      "    if right:\n",
      "        ax.yaxis.tick_right()\n",
      "        \n",
      "pd.set_option('display.width', 500)\n",
      "pd.set_option('display.max_columns', 100)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 1
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "### Description of the data set\n",
      "\n",
      "The data set has been extracted from the Yelp Phoenix restaurants dataset. It is available [here](https://dl.dropboxusercontent.com/u/75194/bigdf.csv)."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "fulldf=pd.read_csv(\"bigdf.csv\")\n",
      "fulldf.head(2)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "html": [
        "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
        "<table border=\"1\" class=\"dataframe\">\n",
        "  <thead>\n",
        "    <tr style=\"text-align: right;\">\n",
        "      <th></th>\n",
        "      <th>user_id</th>\n",
        "      <th>business_id</th>\n",
        "      <th>date</th>\n",
        "      <th>review_id</th>\n",
        "      <th>stars</th>\n",
        "      <th>usefulvotes_review</th>\n",
        "      <th>user_name</th>\n",
        "      <th>categories</th>\n",
        "      <th>biz_name</th>\n",
        "      <th>latitude</th>\n",
        "      <th>longitude</th>\n",
        "      <th>business_avg</th>\n",
        "      <th>business_review_count</th>\n",
        "      <th>user_avg</th>\n",
        "      <th>user_review_count</th>\n",
        "    </tr>\n",
        "  </thead>\n",
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
        "      <td> rLtl8ZkDX5vH5nAx9C3q5Q</td>\n",
        "      <td> 9yKzy9PApeiPPOUJEtnvkg</td>\n",
        "      <td> 2011-01-26 00:00:00</td>\n",
        "      <td> fWKvX83p0-ka4JS3dc6E5A</td>\n",
        "      <td> 5</td>\n",
        "      <td> 5</td>\n",
        "      <td>    Jason</td>\n",
        "      <td> [Breakfast &amp; Brunch, Restaurants]</td>\n",
        "      <td> Morning Glory Cafe</td>\n",
        "      <td> 33.390792</td>\n",
        "      <td>-112.012504</td>\n",
        "      <td> 3.87156</td>\n",
        "      <td> 109</td>\n",
        "      <td> 3.796954</td>\n",
        "      <td> 197</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
        "      <td> SBbftLzfYYKItOMFwOTIJg</td>\n",
        "      <td> 9yKzy9PApeiPPOUJEtnvkg</td>\n",
        "      <td> 2008-05-04 00:00:00</td>\n",
        "      <td> DASdFe-g0BgfN9J2tanStg</td>\n",
        "      <td> 5</td>\n",
        "      <td> 1</td>\n",
        "      <td> Jennifer</td>\n",
        "      <td> [Breakfast &amp; Brunch, Restaurants]</td>\n",
        "      <td> Morning Glory Cafe</td>\n",
        "      <td> 33.390792</td>\n",
        "      <td>-112.012504</td>\n",
        "      <td> 3.87156</td>\n",
        "      <td> 109</td>\n",
        "      <td> 3.473684</td>\n",
        "      <td>  57</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
        "</div>"
       ],
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 2,
       "text": [
        "                  user_id             business_id                 date               review_id  stars  usefulvotes_review user_name                         categories            biz_name   latitude   longitude  business_avg  business_review_count  user_avg  user_review_count\n",
        "0  rLtl8ZkDX5vH5nAx9C3q5Q  9yKzy9PApeiPPOUJEtnvkg  2011-01-26 00:00:00  fWKvX83p0-ka4JS3dc6E5A      5                   5     Jason  [Breakfast & Brunch, Restaurants]  Morning Glory Cafe  33.390792 -112.012504       3.87156                    109  3.796954                197\n",
        "1  SBbftLzfYYKItOMFwOTIJg  9yKzy9PApeiPPOUJEtnvkg  2008-05-04 00:00:00  DASdFe-g0BgfN9J2tanStg      5                   1  Jennifer  [Breakfast & Brunch, Restaurants]  Morning Glory Cafe  33.390792 -112.012504       3.87156                    109  3.473684                 57"
       ]
      }
     ],
     "prompt_number": 2
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The data frame is a frame of reviews. We have joined in information about users and businesses into this frame so that you have only one frame to work with."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "This information is for the reviews themselves:"
     ]
    },
    {
     "cell_type": "raw",
     "metadata": {},
     "source": [
      "  'stars': (star rating, integer 1-5),\n",
      "  'date': (date, formatted like '2011-04-19'),\n",
      "  'review_id': (unique id for the review)."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Here is a description of the data fields in this dataframe, on the business side"
     ]
    },
    {
     "cell_type": "raw",
     "metadata": {},
     "source": [
      "  'business_id': (a unique identifier for this business),\n",
      "  'biz_name': (the full business name),\n",
      "  'latitude': (latitude),\n",
      "  'longitude': (longitude),\n",
      "  'business_review_count': (review count for the restaurant[this is a repeated field for all reviews of the restaurant]),\n",
      "  'categories': [(localized category names)],\n",
      "  'business_avg': (average stars over all users reviews for business[this is a repeated field for all reviews of the restaurant])."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "And Finally, a set of fields for users"
     ]
    },
    {
     "cell_type": "raw",
     "metadata": {},
     "source": [
      "  'user_id': (unique user identifier),\n",
      "  'user_name': (first name, last initial, like 'Matt J.'),\n",
      "  'user_review_count': (count of restaurants reviewed by user[this is a repeated field for all reviews by the user]),\n",
      "  'user_avg': (floating point average of users reviews over all businesses, like 4.31[this is a repeated field for all reviews by the user])."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "In this data set, every user has only one review for each restaurant. Convince yourself of this. (This answer does not need to be submitted)."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Our Recommender\n",
      "\n",
      "To motivate our recommendation system, consider the follwing example. Let's pretend we are in Boston for a second. Lets say the average rating of restaurants here by all the users is 3.5. Sandrine's at Harvard square is better than an average restaurant, so it tends to be rated 0.5 stars above the average (over all the users). However, you are a curmudgeon, who tends to rate 0.2 stars below the average. Then a baseline estimate for the recommendation for Sandrine's, for you, is 3.5+0.5-0.2=3.8.\n",
      "\n",
      "These baseline estimates thus adjust the data by accounting for the systematic tendencies for some users who give higher ratings than others, and for some restaurants to recieve higher ratings than others. We can write the baseline estimate $\\hat Y_{um}^{baseline}$ for an unknown rating $Y_{um}$ for user $u$ and restaurant or business $m$ as:\n",
      "\n",
      "$$ \\hat Y_{um}^{baseline} = \\hat \\mu + \\hat \\theta_{u0} + \\hat \\gamma_{m0} $$\n",
      "\n",
      "where the unknown parameters $\\theta_{u0}$ and $\\gamma_{m0}$ indicate the deviations, or biases, of user $u$ and item $m$, respectively, from some intercept parameter $\\mu$. (The reason for the strange notation with 0s will become clear in Problem 3)\n",
      "\n",
      "Notice that the $\\theta_{u0}$ and $\\gamma_{m0}$ are parameters which need to be fit. The simplest thing to start with, and something we will do for Problems 1 and 2 (but not 3), is to replace them by their \"mean\" estimates from the data. Thus:\n",
      "\n",
      "$$ \\hat Y^{baseline}_{um} = \\bar Y + (\\bar Y_u - \\bar Y) + (\\bar Y_m - \\bar Y)$$\n",
      "\n",
      "where $\\bar Y_u$ =  `user_avg`, the average of all a user $u$'s ratings and $\\bar Y_m$ = `business_avg`, the average of all ratings for a restaurant $m$. $\\bar Y$ is the average rating over all reviews.\n",
      "\n",
      "The final two terms correspond to the user-specific and item-specific bias in ratings, that is, how their ratings tend to systematically diverge from the global average. This is the simplest possible way to predict a rating, based only on information about *this* user and *this* restaurant.\n",
      "\n",
      "Can we do a better job of predicting the rating $Y_{um}$ user $u$ would give to restaurant $r$? According to the central dogma of CF, we ought to be able to use the responses of *similar* users regarding *similar* restaurants to get a better prediction. \n",
      "\n",
      "We can make an estimate of $Y_{um}$ as:\n",
      "\n",
      "$$ \\hat{Y_{um}} = \\hat Y_{um}^{baseline}\\, + \\,\\frac{\\sum\\limits_{j \\in S^{k}(m)} s_{mj} ( Y_{uj} - \\hat Y_{uj}^{baseline} )}{\\sum\\limits_{j \\in S^{k}(m)} s_{mj} } $$\n",
      "\n",
      "where $s^{k}(m)$ is the $k$ neighbor items of item $m$ based on some pooling criterion, for example, those items which have been rated by user $u$.\n",
      "\n",
      "In the next two problems, we will focus on using similar restaurants, or the item neighborhood.\n",
      "To do this, we compute a *similarity measure* $s_{mj}$ between the $m$th and $j$th items. This similarity might be measured via [cosine similarity](http://en.wikipedia.org/wiki/Cosine_similarity), [pearson co-efficient](http://en.wikipedia.org/wiki/Pearson_product-moment_correlation_coefficient) or using other distance based measures. Here we shall use the Pearson coefficient. This measures the tendency of users to rate items similarly. Since most ratings are unknown, it is computed on the \"common user support\" (`n_common`), which is the set of common raters of both items. \n",
      "\n",
      "In the first problem we shall set $S$ to the global neighborhood of the item, and in the second we shall set it to those items which have been rated by user $u$."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "##Q1. Writing a simple \"global\" recommender"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Now we have a way to pool information between similar restaurants to try to predict a user's recommendation. But how do we choose the neighborhood to pool over? We begin with the simplest choice. We calculate the similarity between items using their entire common user support, and rank the nearest neighbors of an item by this similarity. We call this a \"global\" recommender because it assumes that every user perceives the similarity between restaurants in the same way. Later on, we will implement a more specific recommender that pools information based on which items seem the most similar *to this user*.\n",
      "\n",
      "The global recommender does have the advantage of dealing with the possible sparsity of the user's rated items, but also the disadvantage of giving one answer for all users, without taking the user's preferences into account. This is a classic case of bias-variance tradeoff.\n",
      "\n",
      "Lets implement this simpler global recommender first.\n"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "### Exploratory Data Analysis"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**1.1** Visualize the sparsity of the full data set by plotting two histograms of the review count  grouped by the `user_id` and `business_id` respectively. Are there more users or more businesses? "
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "urc=fulldf.groupby('user_id').review_id.count()\n",
      "ax=urc.hist(bins=50, log=True)\n",
      "remove_border(ax)\n",
      "plt.xlabel(\"Reviews per user\")\n",
      "plt.grid(False)\n",
      "plt.grid(axis = 'y', color ='white', linestyle='-')\n",
      "plt.title(\"Review Count per User\");"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAGRCAYAAABbtdZEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nALAQ4QsAAMBChC8AAAALEb4AAAAs9P8BGQfcqm3TAsYAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10ba65910>"
       ]
      }
     ],
     "prompt_number": 3
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "brc=fulldf.groupby('business_id').review_id.count()\n",
      "ax=brc.hist(bins=50, log=True)\n",
      "remove_border(ax)\n",
      "plt.xlabel(\"Reviews per restaurant\")\n",
      "plt.grid(False)\n",
      "plt.grid(axis = 'y', color ='white', linestyle='-')\n",
      "plt.title(\"Review Count per Restaurant\");"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAGRCAYAAABbtdZEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nINBEvvSzAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x107865950>"
       ]
      }
     ],
     "prompt_number": 4
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "print \"Number of Reviews\",fulldf.shape[0]\n",
      "print \"Number of Users\", fulldf.user_id.unique().shape[0], \"Number of Businesses\", fulldf.business_id.unique().shape[0]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Number of Reviews 149319\n",
        "Number of Users 34789 Number of Businesses 4503\n"
       ]
      }
     ],
     "prompt_number": 5
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*your answer here*: There are more users than businesses."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**1.2** Compute the average rating of reviews in the data set and a histogram of all the ratings in the dataset."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "print \"Mean stars over all reviews:\",fulldf.stars.mean()\n",
      "stars=fulldf.stars\n",
      "ax=stars.hist(bins=5)\n",
      "remove_border(ax)\n",
      "plt.xlabel(\"Star rating\")\n",
      "plt.grid(False)\n",
      "plt.grid(axis = 'y', color ='white', linestyle='-')\n",
      "plt.title(\"Star ratings over all reviews\");"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Mean stars over all reviews: 3.74141268023\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmsAAAGRCAYAAAA6rfQGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xl0VFW6xuG3MnCJSBLAMEsGJIDNoOCVQYREUQQMrQFp\nFBpEGoUoYGNrgwrGOIED4gWbSRpE4KJNSxDiRCSG6doRlEFIIpNAmCKEiCQMSWrfP2iqKQMkYkFt\nKr9nLdai9t516tvnW8q7zqnBYYwxAgAAgJX8vF0AAAAAzo+wBgAAYDHCGgAAgMUIawAAABYjrAEA\nAFiMsAYAAGAxwhpwBZgwYYJatWql2NhYhYWFyc/PT3369PF2WeXWunVrvfXWW94uw1orV67UY489\npsjISElScXGxpk2bpnvuuUePPfbYZa+HfgF2IawBlluwYIGef/55ffzxx0pLS9OePXvUp08f7d27\n123drl27vFRhab+s5eabb1Z4eLiXqrFfhw4dFBYW5jpvAQEBGjhwoDIzM1VYWHjZ66FfgF0Ia4Dl\nFi1apMjISNWuXVuSVLlyZb377ruqVKmSa40xRgMHDvRWiW6ysrI0btw4t7EpU6bonnvu8VJF9nM4\nHKXCUaVKlVS3bl2v1EO/ALsQ1gDLFRUV6bvvvlN6erprLDAw0C2cvfDCC/ryyy+9UJ27o0ePqk+f\nPjpx4kSpOafT6YWKcLHoF2APwhpguX79+snpdKpr16566623XP+I9uvXT5K0Z88effXVV5KkJ598\nUu+++64kafPmzRo6dKimT5+u++67T1OmTJEk/fzzz3rnnXfUrl07/etf/9JNN92khg0blvrHec+e\nPXrxxRd12223aeXKlapbt64GDBggSZo+fbrGjh2riRMn6s4779SWLVskSampqcrLy9PatWv15JNP\nasuWLUpJSdG9996roUOHSpJ++OEH/fWvf1WzZs20adMm3XLLLQoODtbjjz/u9vo//vijhg0bpuee\ne0716tWTn5+fOnXqpHnz5kmS5syZo9dff10TJkxQrVq1tGHDhvOew+zsbD3yyCN6/vnnFR8fr/vu\nu891G3nDhg1q2LCh/P39XYH46NGjuueee9S7d2+dOnVKkjRt2jSNGDFCd911l9q0aaO1a9dKktLS\n0tS/f3+99tprevLJJ1W1alV9/vnnpWowxujFF1/U+PHjNW7cOHXr1k379u0ro/vnd6E+7tq1SyNH\njtRDDz2kZs2a6amnnpLT6VRubq7uvvtu+fn5KSkpyXWsxMRENW/eXLt379bSpUvd+iVJJ06c0HPP\nPaeEhATdfPPNio+P148//ihJGjx4sPz8/DR8+HA5nU5NnjzZ1avvv/9ekvTRRx+pZs2a2r17tyTp\n2Wef1YwZM/TUU08pLCzsos8BUGEYANabOHGiqVy5snE4HKZ169Zm06ZNbvOzZs0yDofDbeyGG24w\nL730kjHGmG+++cb4+/ub3bt3m7y8PDNx4kTjcDhMYmKiSU9PN8OGDSv1mvv37zd/+MMfTPXq1c3c\nuXPN3LlzzeTJk8369euNw+Ewe/fuNcYYM3z4cHP77be7nhcTE2MGDhzoerxv3z4THR3tGispKTGv\nvfaaCQwMNBMmTDCFhYVm4cKFxuFwmPXr17ue17VrVzNnzhxjjDErVqwwDofDzJo1yxhjzPHjx03t\n2rVdaz/88EO3555t3759platWua7775zjfXu3ds0bNjQHDt2zBhjzPvvv28cDofJzMx0rUlISDAH\nDx40xhgzb94887e//c01161bN1O3bl1TXFxsNm7caIKDg01MTIz56quvzBNPPOH2WmckJyebSpUq\nuR736NHDDBo0yPX4XD385bk82/n6WFJSYrp27WqOHz9ujDEmIyPDOBwO8/bbbxtjjMnJyTGBgYFm\nypQprmPNnTvXLFq0yHW+zu6XMcY88sgjZvPmzcYYYwoLC80111xj7rvvPmOMMfn5+eaqq64yb731\nllvdDzzwgOvx1q1bzTPPPGOMMeaLL74w8fHxrrmxY8eec38A/iPA22ERQNlGjBihbt26KSEhQV98\n8YX++7//Wx988IHi4uLO+5xBgwapQ4cOkqSrrrrKdcWlQ4cOatmypSTp/vvvV3R0tDp27Fjq+bVr\n11aTJk2Unp6uvn37usaPHDmi0aNHq2bNmq5j79y50zVvjHE7Tp06ddzee+Xn56drrrlGxcXF+vOf\n/yxJrn1kZWWpZcuWKigo0GeffaaRI0dKkm699VbVq1fPdTWnqKhIhw8f1ttvv61HH31UcXFxOnbs\n2DnPw9tvv63q1avrd7/7nWts7Nixat68uebMmaOhQ4cqPj5edevW1axZszR+/HidOHFCJ06ccO0x\nKSlJMTExGj16tCQpNDRUDRs2VG5urpo3b67q1aurXbt2atOmjdq0aXPOOlq0aKGxY8e6Hv/yvP1a\n1apVO2cfFyxYoN27d+v55593rb311lt15MgRSVK9evUUHx+vWbNmaciQIZKkzz//XLNmzZJUul97\n9+7VBx98oGrVqrnG2rRp47oSGxISovj4eL333nsaPny4JOmmm27StGnTdPz4cQUFBWnevHkaNGiQ\nJOnkyZNKTU3VV199pbZt2+rRRx+96HMAVBSENeAK0ahRIy1btkzjx4/X6NGj1bdvX23fvv28t5Ee\ne+wxbd++Xa+//rrrH9Zf3uqsXLlyma/7yzXVqlXTSy+9pI8++kjbtm3T1q1b3Y7rcDh+7dZcH5Y4\nefKkpNNhzBijHTt2uNbUr19fUVFRkqSqVavq+eef17Bhw5SSkqIpU6ac99OL69atU5UqVdzGrr/+\nelWqVEnr16+XdPrTl4MGDdL06dP18ssva/HixerVq5ckqbCwUFu3btXSpUt13XXXnXcPZZ3LyMhI\njR49WvPnz1dubq4OHDhwUeeqrNf+9ttv1bJlS73yyivnXf/II4/o9ttv1+bNm1WtWjXVrVtXfn7/\neVfM2YF748aNCgoKuuDxBg0apNtuu01btmxRo0aN9P333yswMFALFy5U3759tXPnTtfXktx1111q\n3769br31Vg0bNkwvvvjib9k6UCHwnjXAYrt379aaNWvcxv76179q4MCBOnbsmFavXn3e506ZMkXD\nhw/XY489pt69e3uspsLCQnXp0kUHDx7UyJEj1axZM48d+4zQ0FDdd999mjFjhk6dOqXCwkIFBAS4\nXUkcPXq0Fi5cqE2bNqlFixalztMZ/v7+2rNnj9uYw+FQ9erVFRgY6BobPHiwDh06pJSUFH3xxRe6\n6667JEnHjx8vFRzPOPN+tvLIzc1V27ZtVaNGDT3++OOKiIgodRXSEwoLC895xa6oqMj199jYWEVH\nR2vmzJmaP39+qU8Snx0iCwsLlZubq+PHj5/3mDExMYqKitLs2bP1wQcfaMCAAfrDH/6gOXPm6JNP\nPlHXrl3djr1kyRIlJiZq2rRpat26tQ4dOvSb9w34MsIaYLHg4GA9++yzpcZbtWolSapVq5ak0lez\ncnJyNGzYMD3yyCOqXLmyRz/Z99ZbbykjI0ODBw+WdO5PDXoihMyYMUN16tTRqFGjNHPmTCUnJ7uu\nwOXm5mrTpk2Kj49XZmamWrRooddee+2cx2nXrp1yc3O1bds211hRUZEOHTqk9u3bu8bq16+vbt26\nafz48apXr57rnNaoUUPVq1fX9OnT3Y67YcMGLVu2rNz7GTNmjIqKitSlSxdJUklJSbmf+2tER0cr\nIyNDGzdudBv/5fl55JFHNHfuXH3//feKjo4+7/EaNWqkkpISzZw502181qxZbiHrwQcf1Ny5c7Vk\nyRLFx8frwQcfVFpamqZMmaL4+HjXuhUrVkiSnnnmGX377bfKy8tzfWgEwLn9qrD2ww8/6NVXX9Xs\n2bNd7x0BcOmEhoZq/fr1euKJJ1z/uBcXFys5OVk333yz6/1R1atXl3T6PV/r169Xbm6unE6nMjIy\ndPLkSf3jH/+QdPr9R3l5ea6AdfbVlnNxOp0qLi52G9u3b58KCgq0ZcsW7d+/X2lpaTpy5IgOHz6s\nU6dOqUaNGsrKypIxRt9++62k01egzr4K9ctjnpk7e489evRQXFyc2rRpo5o1a2r16tXKz8+XdPpq\nz9SpUyVJV199tXr16qV69eqdcw9Dhw5V3bp19eqrr7rG3n//fTVv3rzUFcchQ4boq6++Uv/+/d3G\nExIS9OGHH2rIkCFauXKl5syZo5deekndu3d31V3Wudy/f7/27Nmj/fv3a+vWrVq7dq0OHjzo+n/p\nmXNydvgtKioqda7Odq4+9uvXT1dffbXi4uL0/vvv68svv9RDDz3kCvhnPPjggzp27JhbYD3j7H61\naNFCHTp00JNPPqk333xTq1at0iuvvKJdu3apTp06bsfLzc1VTEyMpNNfrBsdHa1GjRq5XcHcuXOn\nkpOTJZ0Olu3btz9v7wD8W3k/ifD++++bdu3amR07drjGcnJyzNChQ82UKVNM//793T4BdSnmgIqo\ndevWxuFwmIiICHPvvfeaTp06mYceesj8+OOPrjUFBQWmdevWpn79+ubdd981TqfT9OrVywQFBZmO\nHTuaTZs2mVatWpkmTZqYdevWmZ49exo/Pz+TkJBgtmzZcs7XXbFihWnevLnx9/c3kydPNrm5ucYY\nYzZt2mSuu+46ExISYkaMGGHS0tJMaGiouf/++82pU6fMZ599ZkJDQ03Hjh3Njh07zNy5c01wcLC5\n9tprzccff2x27dpl7rjjDuPn52feeOMNc+zYMTN27FjjcDhMt27dzJ49e0xxcbHp0aOHuf76601I\nSIgJCAgwDofD1KtXz+Tl5ZmdO3caf39/M3z4cDN16lTz8MMPm7y8vPOew+3bt5u7777b9O3b14wd\nO9Y8+uij5vDhw6XWOZ1OM3To0FLjp06dMo8++qipVq2aqVGjhhkwYIA5fPiwKS4uNhMnTjT+/v6m\nefPmZunSpeet4YsvvjB169Y1YWFhJikpybz//vuuc5iRkWE6dOhg/Pz8zKuvvmry8vLMzJkzTZUq\nVUxUVJT55JNPzrmn8/VxxYoV5oYbbjCVK1c2zZs3Nx9++OE5axo5cqQpLCx02/8v+2WMMXv27DHd\nunUzQUFB5tprrzVJSUnnPN7QoUPNiRMnXI8nTpxosrKy3NbMnj3bVKtWzbz00ktmwoQJZtSoUec9\nZwBOK1dYS0tLM2FhYa6P6htz+j/qVq1amWXLlhljjNmyZYuJjIw0JSUlHp8rLi726KYB2C07O9uM\nHDnSbaygoMBMmzbtvMEDAHxVmZ8GNcZo6NChGj58uNvHuVNTU5WZmem65N20aVMFBgZq0aJFCg4O\n9uhccnKyevbs6dErigDsNWzYMCUkJLiNXXXVVYqKinL97BYAVBRlhrX/+7//U3Z2tn744Qf16tVL\nmzdv1mOPPaYff/xRkZGRCgj4zyGio6O1fPly1axZ0+NzhDWg4iguLtabb76pqKgoRUdH6+jRo0pN\nTdWBAwfUuXNnb5cHAJdVmWFt3bp1qlq1qsaNG6drrrlG33zzjW6++WbdcccdCgkJcVsbGhqqnJwc\nOZ1Oj82FhIQoJyfnYvcH4Ao0Z84cjRgxQh07dlRRUZFatmyp4cOHu75EFwAqkjLD2rFjx9S4cWNd\nc801kk5/ZcBNN92k6667rtRHw51Op4wxCggIcPv0z2+dO5cHH3xQERERrscxMTGu26cArmz16tXT\nwoULvV0GAFihzLBWu3ZtFRQUuI3Vr19fb7/9tuunTs7Iz89XgwYNVKdOHa1cudJjc2eHsjPefffd\nS/KFkgAAADYp83vW2rVrp927d7t9j8/JkyeVmJio7du3u63NyspSbGysYmNjS33b98XOZWdnc8UM\nAABUWGWGtSZNmqh169ZaunSppNNflrhx40Y9/PDDCg8PV1pamqTTgaugoEBxcXFq27atx+YKCwsv\n+GPVAAAAvqxcP+Q+d+5cPfHEE8rOzlZOTo5mzJih2rVra/HixUpKSlJmZqYyMjKUkpKioKAgSfLY\n3NKlS11zAAAAFY3DXKFv/HI4HLxnDQAA+Dx+yB0AAMBihDUAAACLEdYAAAAsRlgDAACwGGENAADA\nYoQ1AAAAixHWAAAALEZYAwAAsBhhDQAAwGKENQAAAIsR1gAAACxGWAMAALAYYQ0AAMBihDUAAACL\nEdYAAAAsRlgDAACwGGENAADAYoQ1AAAAixHWAAAALEZYAwAAsBhhDQAAwGKENQAAAIsR1gAAACxG\nWAMAALAYYQ0A4DFFzhJvl4Cz0A/f4DDGGG8XcTEcDoeu0NIBwKfVnzXK2yXg33IGjvN2CfAArqwB\nAABYjLAGAABgMcIaAACAxQhrAAAAFiOsAQAAWIywBgAAYDHCGgAAgMUIawAAABYjrAEAAFiMsAYA\nAGAxwhoAAIDFCGsAAAAWI6wBAABYjLAGAABgMcIaAACAxQhrAAAAFiOsAQAAWIywBgAAYLFfHdby\n8vJUWFh4KWoBAADAL5QrrHXo0EF+fn7y8/NT+/btddVVV2nv3r1KSEjQ1KlTNWDAAG3evNm1/lLM\nAQAAVEQOY4y50IJ169bp448/Vvfu3SVJ9evXV1hYmG666SaNHz9enTt3VmZmprp3765t27bJ4XB4\ndG7r1q3y9/cvXbjDoTJKBwB4Qf1Zo7xdAv4tZ+A4b5cADwgoa8HEiRPVokULVa1aVY0aNZIkLVu2\nTJmZmYqJiZEkNW3aVIGBgVq0aJGCg4M9OpecnKyePXt6fOMAAABXggveBi0pKVFeXp7eeOMNNW7c\nWH369FFRUZFWr16tqKgoBQT8J+tFR0dr+fLlWrNmjSIjIz06BwAAUFFd8Mqav7+/UlJSZIzRvHnz\nNHToUD399NM6duyYgoOD3daGhoYqJydHTqdTISEhHpkLCQlRTk7Ob9kfAADAFa3M26DS6feH9evX\nTydOnNCYMWPUq1cvBQYGuq1xOp0yxiggIMCjcxeSmJjo+ntMTIzrFioAwHtG3tDZ2yUAPqVcYe2M\n3//+9xo2bJjq1KmjlStXus3l5+erQYMGHp+LiIg4bz1nhzUAgB0mrE/1dgn4t5E3Epx9wa/6nrWS\nkhI1btxYsbGx2rFjh9tcVlaWYmNjPTqXnZ3N1TIAAFChXTCsff3113rnnXdctyMnTZqkZ555Ru3a\ntVN4eLjS0tIknQ5cBQUFiouLU9u2bT02V1hYqLi4uEu2eQAAANtd8DbogQMHNGbMGM2dO1ddunRR\nmzZt1KNHD0nS4sWLlZSUpMzMTGVkZCglJUVBQUEenVu6dKlrDgAAoCIq80txbcWX4gKAnfhSXHvw\npbi+gR9yBwAAsBhhDQAAwGKENQAAAIsR1gAAACxGWAMAALAYYQ0AAMBihDUAAACLEdYAAAAsRlgD\nAACwGGENAADAYoQ1AAAAixHWAAAALEZYAwAAsBhhDQAAwGKENQAAAIsR1gAAACxGWAMAALAYYQ0A\nAMBihDUAAACLEdYAAPBRRc4Sb5cADwjwdgEAAODSCPTzV/1Zo7xdBiTlDBx30c/lyhoAAIDFCGsA\nAAAWI6wBAABYjLAGAABgMcIaAACAxQhrAAAAFiOsAQAAWIywBgAAYDHCGgAAgMUIawAAABYjrAEA\nAFiMsAYAAGAxwhoAAIDFCGsAAAAWI6wBAABYjLAGAABgMcIaAACAxQhrAAAAFiOsAQAAWIywBgAA\nYDHCGgAAgMUIawAAABYjrAEAAFiMsAYAAGCxcoc1p9Op2NhYpaenS5L27t2rhIQETZ06VQMGDNDm\nzZtday/FHAAAQEUUUN6FU6ZM0caNG+VwOGSMUY8ePTR+/Hh17txZnTp1Uvfu3bVt2zY5HA6Pzm3d\nulX+/v6X8hwAAABYq1xhbdWqVYqMjFRwcLAkKTU1VZmZmYqJiZEkNW3aVIGBgVq0aJGCg4M9Opec\nnKyePXt6dNMAAABXijJvgx4+fFhr1qxRt27dJEnGGK1evVqRkZEKCPhP1ouOjtby5cu1Zs0aj88B\nAABUVGVeWZs4caLGjBnjNnbw4EGFhIS4jYWGhionJ0dOp9NjcyEhIcrJyflVGwIAAPAlFwxrM2bM\nUN++fVWpUiW3cX9/fwUGBrqNOZ1OGWMUEBDg0bkLSUxMdP09JibGdQsVAOA9I2/o7O0ScBb6ceUr\nM6wNHz7c9fjkyZO68847ZYzR7373O7e1+fn5atCggerUqaOVK1d6bC4iIuK89Z0d1gAAdpiwPtXb\nJeDfRt7YmX5YYuSNFx+aL/ietYyMDB0/ftz1Jzw8XMuWLVN6erq2b9/utjYrK0uxsbGKjY3Vjh07\nPDKXnZ3N1TIAAFChXdSX4rZt21bh4eFKS0uTdDpwFRQUKC4uzqNzhYWFiouL88Q+AQAArkjl/p61\nszkcDi1evFhJSUnKzMxURkaGUlJSFBQUJEkem1u6dKlrDgAAoCJyGGOMt4u4GGe+nBcAYJf6s0Z5\nuwT8W87AcfTDEjkDx130c/ltUAAAAIsR1gAAACxGWAMAALAYYQ0AAMBihDUAAACLEdYAAAAsRlgD\nAACwGGENAADAYoQ1AAAAixHWAAAALEZYAwAAsBhhDQAAwGKENQAAAIsR1gAAACxGWAMAALAYYQ0A\nAMBihDUAAACLEdYAAAAsRlgDAACwGGENAADAYoQ1AAAAixHWAAAALEZYAwAAsBhhDQAAwGKENQAA\nAIsR1gAAACxGWAMAALAYYQ0AAMBihDUAAACLEdYAAAAsRlgDAACwGGENAADAYoQ1AAAAixHWAAAA\nLEZYAwAAsBhhDQAAwGKENQAAAIsR1gAAACxGWAMAALAYYQ0AAMBihDUAAACLEdYAAAAsRlgDAACw\nGGENAADAYuUKa99++61uueUWVatWTXfccYcOHz4sSdq7d68SEhI0depUDRgwQJs3b3Y951LMAQAA\nVDRlhrVTp07pH//4h1JTU5WTk6Njx45pwoQJkqQePXooPj5eQ4YM0ahRoxQXFyen0yljjEfnSkpK\nLvmJAAAAsFFAWQuOHDmixMREVapUSZLUqVMn+fv7a9myZcrMzFRMTIwkqWnTpgoMDNSiRYsUHBzs\n0bnk5GT17NnT45sHAACwXZlX1mrVquUKaidPntTBgwf1+OOPa/Xq1YqKilJAwH/yXnR0tJYvX641\na9YoMjLSo3MAAAAVUZlX1s5YsmSJnn32WeXl5Wnz5s06cOCAgoOD3daEhoYqJydHTqdTISEhHpkL\nCQlRTk7Or90XAACATyj3p0Hj4uK0ePFidezYUf369VNgYKACAwPd1px531lAQIBH5wAAACqqcl9Z\nk6SIiAjNnDlTNWrUUFhYmH766Se3+fz8fDVo0EB16tTRypUrPTYXERFxznoSExNdf4+JiXG91w0A\n4D0jb+js7RJwFvpx5XMYY8yvfVKDBg20YMEC3XXXXTp69KhrvGHDhnrllVd07bXXqkuXLh6d6927\nt3vhDocuonQAwCVWf9Yob5eAf8sZOI5+WCJn4LiLfm6Zt0Hz8vK0ZMkS1+P09HT1799f7du3V3h4\nuNLS0iRJWVlZKigoUFxcnNq2beuxucLCQsXFxV30BgEAAK5kZd4G3bFjhwYPHqzGjRurV69euvrq\nq/Xiiy9KkhYvXqykpCRlZmYqIyNDKSkpCgoK8ujc0qVLXXMAAAAVzUXdBrUBt0EBwE7cdrMHt0Ht\ncUlvgwIAAMB7CGsAAAAWI6wBAABYjLAGAABgMcIaAACAxQhrAAAAFiOsAQAAWIywBgAAYDHCGgAA\ngMUIawAAABYjrAEAAFiMsAYAAGAxwhoAAIDFCGsAAAAWI6wBAABYjLAGAABgMcIaAACAxQhrAAAA\nFiOsAQAAWIywBgAAYDHCGgAAgMUIawAAABYjrAEAAFiMsAYAAGAxwhoAAIDFCGsArnhFzhJvlwAA\nl0yAtwsAgN8q0M9f9WeN8nYZkJQzcJy3SwB8DlfWAAAALEZYAwAAsBhhDQAAwGKENQAAAIsR1gAA\nACxGWAMAALAYYQ0AAMBihDUAAACLEdYAAAAsRlgDAACwGGENAADAYoQ1AAAAixHWAAAALEZYAwAA\nsBhhDQAAwGKENQAAAIsR1gAAACxGWAMAALBYmWEtPT1dLVu2VHBwsLp06aI9e/ZIkvbu3auEhARN\nnTpVAwYM0ObNm13PuRRzAAAAFVHAhSZzc3P197//XfPmzdPevXv1yCOP6KGHHtKyZcvUo0cPjR8/\nXp07d1anTp3UvXt3bdu2TQ6Hw6NzW7dulb+//+U6HwAAAFa5YFhbvny5Jk+erKpVq6pZs2ZKTEzU\n0KFDlZqaqszMTMXExEiSmjZtqsDAQC1atEjBwcEenUtOTlbPnj0v1f4BAACsdsGw1qdPH7fHtWrV\nUoMGDbR69WpFRkYqIOA/T4+Ojtby5ctVs2ZNj88R1gAAQEV1wbD2S998842GDh2q7OxshYSEuM2F\nhoYqJydHTqfTY3MhISHKycn5NSUCAAD4lHKHtYKCAm3atEnz5s3TiBEjFBgY6DbvdDpljFFAQIBH\n5y4kMTHR9feYmBjXLVQAFc/IGzp7uwT8G72wC/248pU7rL3++uuaNGmS/P39VbduXa1atcptPj8/\nXw0aNFCdOnW0cuVKj81FRESct6azwxqAim3C+lRvlwBJI2/sTC8sQj/sMfLGiw/N5fqetRkzZqhf\nv34KCwuTJHXo0EE7duxwW5OVlaXY2FjFxsZ6bC47O5urZQAAoEIrM6zNnj1bQUFBKioqUlZWltLT\n07Vjxw5FREQoLS1N0unAVVBQoLi4OLVt21bh4eEemSssLFRcXNyl2jsAAID1Lngb9NNPP9XgwYNV\nUlLiGnM4HMrOzlbHjh2VlJSkzMxMZWRkKCUlRUFBQZKkxYsXe2Ru6dKlrjkAAICKyGGMMd4u4mI4\nHA5doaUDuATqzxrl7RIgKWfgOHphEfphj5yB4y76ufw2KAAAgMUIawAAABYjrAEAAFiMsAYAAGAx\nwhoAAIDFCGsAAAAWI6wBAABYjLAGAABgMcIaAACAxQhrAAAAFiOsAQAAWIywBgAAYDHCGgAAgMUI\nawAAABZYDBGBAAASEUlEQVQjrAEAAFiMsAYAAGAxwhoAAIDFCGsAAAAWI6wBAABYjLAGAABgMcIa\nAACAxQhrAAAAFiOsAQAAWIywBgAAYDHCGgAAgMUIa8BFKnKWeLsEAEAFEODtAoArVaCfv+rPGuXt\nMiApZ+A4b5cAAJcMV9YAAAAsRlgDAACwGGENAADAYoQ1AAAAixHWAAAALEZYAwAAsBhhDQAAwGKE\nNQAAAIsR1gAAACxGWAMAALAYYQ0AAMBihDUAAACLEdYAAAAsRlgDAACwGGENAADAYoQ1AAAAixHW\nAAAALParwtqJEyd09OjRS1ULAAAAfqFcYc0Yo9mzZys6Olpff/21a3zv3r1KSEjQ1KlTNWDAAG3e\nvPmSzgEAAFQ0AeVZdOjQIXXu3FkPPfSQHA6HpNMBrkePHho/frw6d+6sTp06qXv37tq2bZscDodH\n57Zu3Sp/f/9LeiIAAABsVK6wFhYWVmosNTVVmZmZiomJkSQ1bdpUgYGBWrRokYKDgz06l5ycrJ49\ne/7mzQIAAFxpyhXWzmX16tWKiopSQMB/DhEdHa3ly5erZs2aioyM9OgcYQ0AAFREFx3WDhw4oODg\nYLex0NBQ5eTkyOl0KiQkxCNzISEhysnJudgyfcrRU8f1f/t3aH/hT94upcK7uVaErq9e19tlAAAq\ngIsOawEBAQoMDHQbczqdMsZ4fO58EhMTXX+PiYlx3T71VZX8AnS06ITyThR6u5QK73hxkSRp5A2d\nvVwJzqAX9qAXdqEfV76LDmt169bVqlWr3Mby8/PVoEED1alTRytXrvTYXERExDlrODusVQQnS4q1\ncNs3Wr1/u7dLqfCuDvwvta4ZrgnrU71dCiSNvLEzvbAEvbAL/bDHyBsvPjRf9JfixsTEaMeOHW5j\nWVlZio2NVWxsrMfmsrOzff6KGQAAwPmUO6yduR1pjJEktWvXTuHh4UpLS5N0OnAVFBQoLi5Obdu2\n9dhcYWGh4uLiPLdjAACAK0i5boP++OOPmjFjhhwOh+bPn6969eqpSZMmWrx4sZKSkpSZmamMjAyl\npKQoKChIkjw2t3TpUtccAABAReMwZy6VXWEcDoeu0NIv2k8nj+vhtLm8Z80CY/+7ux5udqvqzxrl\n7VIgKWfgOHphCXphF/phj5yB4y76ufyQOwAAgMUIawAAABYjrAEAAFiMsAYAAGAxwhoAAIDFCGsA\nAAAWI6wBAABYjLAGAABgMcIaAACAxQhrAAAAFiOsAQAAWIywBgAAYDHCGgAAgMUIawAAABYjrAEA\nAFiMsAYAAGAxwhoAAIDFCGsAAAAWI6wBAABYjLAGAABgMcIaAACAxQhrAAAAFiOsAQAAWIywBgAA\nYDHCGgAAgMUIawAAABYjrAEAAFiMsAYAAGAxwhoAAIDFCGsAAAAWI6wBAABYjLAGAABgMcIaAACA\nxQhrAAAAFiOsAQAAWIywBgAAYDHCGgAAgMUIawAAABYjrAEAAFiMsAYAAGAxwhoAAIDFCGsAAAAW\nI6wBAABYjLAGAABgMcIaAACAxawLa3v37lVCQoKmTp2qAQMGaPPmzd4uCQAAwGusCmvGGPXo0UPx\n8fEaMmSIRo0apbi4OJWUlHi7NGvkbsz2dglecTJrt7dL8Ar2XbGw74qFfVcsX3755UU/16qwlpqa\nqszMTMXExEiSmjZtqsDAQCUnJ3u3MIsQ1ioW9l2xsO+KhX1XLD4T1lavXq2oqCgFBAS4xqKjo7V8\n+XIvVgUAAOA9AWUvuXwOHDig4OBgt7GQkBDl5OR4qSL7RIfW0o2/u9XbZVx2aWt2K9aifbe8pr63\nSwAAVBAOY4zxdhFnPPbYY9q0aZPS09NdYw888IAKCgq0ePFit7XXXXedtm/ffrlLBAAA+NUGDBig\n2bNnX9RzrbqyVrduXa1atcptLD8/XxEREaXWbtu27TJVBQAA4D1WvWctNjZWO3bscBvLzs52feAA\nAACgorEqrLVt21bh4eFKS0uTJGVlZamwsFBxcXFergwAAMA7rAprDodDixcv1rvvvqu//e1vGjdu\nnJYuXaqgoKBfdZy9e/deogovnxMnTujo0aPeLuOy+7X79oVe49eh5xUL/a5YfKnfeXl5Kiws9Mix\nrAprkhQVFaXZs2crISFBL730kmbOnFnmrxmkpqbKz8/P9WfFihWXuWrPMcZo9uzZio6O1tdff33e\nddOnT1dSUpKef/55jRkz5jJWeGmUd9++1GtJSk9PV8uWLRUcHKwuXbpoz54951zna/2Wyr93X+r5\nt99+q1tuuUXVqlXTHXfcocOHD59znS/2u7x796V+n+F0OhUbG+v24bmz+WK/zyhr777W7w4dOrj2\n0r59e1111VWl1lxUv42lnE6nadWqlVm2bJkxxpgtW7aYyMhIU1xcXGrtkCFDzLp168y6devMhg0b\nLnepHpWbm2v27NljHA6H+eKLL865Jjk52bRv3971uHfv3uadd965XCVeEuXZtzG+1euDBw+a/v37\nm02bNplPP/3UhIeHm86dO5da54v9Lu/ejfGdnp88edKMHj3aFBYWmmPHjpm2bduap59+utQ6X+x3\nefdujO/0+2yTJ0821atXN+np6aXmfLHfZ7vQ3o3xrX6vXbvWJCUlufZz8ODBUmsutt/WXVk7o7y/\nZrB161Zt2rRJ+/btU7NmzdSiRQsvVOs5YWFhql//wt/h9eqrr6pr166ux/fcc48mTpx4qUu7pMqz\nb1/r9fLlyzV58mQ1a9ZMXbp0UWJiYqlPQ0u+2e/y7t2Xen7kyBElJiYqKChIVapUUadOneTv719q\nnS/2u7x796V+n7Fq1SpFRkaW+g7RM3yx32eUtXdf6/fEiRNVuXJlVa1aVa1atVLNmjVLrbnYflsb\n1sr7awbr1q3T8ePHde+99+raa69Vamrq5S71sjp16pTWrl2rJk2auMYaNWqkzZs369ChQ16s7NLz\ntV736dNHVatWdT2uVauWwsPD3db4ar/Ls3fJt3peq1YtVapUSZJ08uRJHTx4UH/+85/d1vhqv8uz\nd8m3+i1Jhw8f1po1a9StW7dzzvtqv6Wy9y75Vr9LSkqUl5enN954Q40bN1afPn1UVFTktua39Nva\nsFbeXzPo06eP1q1bp507d+qmm25SfHy8Dhw4cDlLvazy8vJUVFSkkJAQ11hoaKgk+fwvPfh6r7/5\n5hsNGTLEbayi9Ptce5d8s+dLlixRmzZtlJqaqu+++85tztf7faG9S77X74kTJ+rxxx8/77wv97us\nvUu+1W9/f3+lpKRo//79mjNnjlJSUvT000+7rfkt/bY2rAUEBCgwMNBtzOl0nnd9/fr1tXDhQtWu\nXbvUrx34kjNXGs8+N2fOi7HnxyguKV/sdUFBgTZt2qThw4e7jVeEfp9v72fzpZ7HxcUpOTlZHTt2\nVL9+/dzmfL3fF9r72Xyh3zNmzFDfvn1dVxSl0j301X6XZ+9n84V+n+FwONSvXz+9+eabmjt3rtvc\nb+m3tWGtbt26+umnn9zG8vPzVa9evfM+JygoSHfeeafy8/MvdXleU6NGDQUGBrqdmzP7vdC58TW+\n1uvXX39dkyZNkp+f+3+SFaHf59v7L/lSzyMiIjRz5kwdOnTI7VORFaHf59v7L13p/Z4xY4ZuvPFG\nBQUFKSgoSLt27dKdd96pPn36uNb4ar/Ls/dfutL7/Uu///3vS+3lt/Tb2rB2sb9mUFJS4nY/2Nc4\nHA7FxMRo69atrrGsrCw1bdr0nG9m9GW+0usZM2aoX79+CgsLkyS39zn4er8vtPdz8ZWeS1LlypVV\no0YNVa9e3TXm6/0+41x7P5crud8ZGRk6fvy46094eLiWLVumBQsWuNb4ar/Ls/dzuZL7/UslJSVq\n3Lix29hv6be1Ye18v2Zw991369lnn9WmTZskSRMmTFBWVpak0+9zy87OVvfu3b1Wtyec67Lo2Xv+\n05/+pCVLlrjmPv74Yz300EOXt8hLoKx9+2KvZ8+eraCgIBUVFSkrK0vp6emaP39+heh3efbuSz3P\ny8tz62N6err69+8vh8Ph8/0u7959qd8X4uv9vhBf7ffXX3+td955x/Xv2KRJk/TMM89I8ky/rfoh\n97Od+TWDpKQkZWZmKiMjQ0uXLtVVV12lTz/9VK1atVKzZs30+eef64UXXtCQIUMUEhKihQsXun2C\n9Erz448/asaMGXI4HJo/f77q1aunJk2auPbcvHlz3Xfffdq1a5eeffZZBQUFKTw8XCNHjvR26b9J\nWfv2xV5/+umnGjx4sEpKSlxjDodDWVlZmjRpkk/3uzx797We79ixQ4MHD1bjxo3Vq1cvXX311Xrx\nxRclyef/+y7P3n2t3xfi6/2+EF/t94EDBzRmzBjNnTtXXbp0UZs2bdSjRw9Jnum3w1zJ72IEAADw\ncdbeBgUAAABhDQAAwGqENQAAAIsR1gAAACxGWAMAALAYYQ0AAMBihDUAAACLEdYAAAAsRlgDYLXM\nzEx17txZDzzwgFq0aCE/Pz99+OGH3i7L5fjx44qKitKaNWu8XQoAH0VYA2C1P/3pT+rTp4/mz5+v\nDRs2aNCgQTp48KBrfsqUKZe9prNfMygoSE888YTP/AA1APsQ1gBYbf369SoqKpJ0+vdD33zzTf38\n88+SpM8++0xvvPHGZa1nw4YNevLJJ93GHn30UVWvXv2y1gGg4iCsAbBaq1atNGrUKH300UeSpKuv\nvloPPvigJCklJUV5eXl6+eWXtX37du3bt08JCQmaNWuWevfurfXr10uS/vd//1cdO3bUhx9+qGuv\nvVZTp051e41PP/1Ud999t9577z01adJEo0ePPu+xUlNTVVhYqJdfflnr1q3TkSNHNG3aNKWnp0uS\nJk6cqGbNmmnJkiVq166dGjdurB9++MH1WosWLdLLL7+se++9V35+foqLi9PWrVsv8VkEcEUzAGCx\n7du3m+uvv944HA4THx9vDh486Jr78ssvTUREhOvx6NGjzaOPPmqMMWbKlCkmPj7eGGPM4cOHjcPh\nMH//+9/Nv/71L7Nhwwa31zh58qQJCwszzz33nMnMzDQrVqw477F27txpHA6H67kpKSmmSpUqZubM\nmcYYY7Kysoyfn59ZsWKFMcaYnj17mjFjxhhjjMnLyzPh4eHGGGOKi4tNnTp1zP/8z/947FwB8E0B\n3g6LAHAhUVFR+vbbbzVhwgQlJSXpxhtv1IoVK9SwYUMZY9zWjhgxQk6nU4cOHdLGjRt15MgRSXLd\norztttsUHh5e6jUqVaqkKlWqKDY2Vk2aNFGTJk0UHR19zmP9Urdu3RQWFuZ6/F//9V8yxujWW2+V\nJDVr1kx79uyRJH3//fc6evSoJMnf31/t27eXv7//bzxDAHwdt0EBWMvpdEo6HaZGjRqlVatWqbCw\nUH/5y1/Oub5GjRqaPn26li1bptatW5cKcw6H44Kvd/Z8Wce6mH00bdpU/v7+rlumRUVF6tq160Uf\nF0DFQFgDYK2PPvpIGzZscD1u1aqV+vfvr507d55z/VNPPSVJuv/++3/zFStPHevsABgcHKz33ntP\nM2fO1MyZMzV27FhFRkb+pjoB+D7CGgBrRUZGatCgQTp27JhrbNeuXbrzzjslSVWqVNGRI0fkdDp1\n8OBBpaamqqioSMYYrVu3Tj/99JNKSkpcV8XOXOE6F2OM2/z5jlWlShVJ0qFDh5SbmytJKikpUUlJ\nyTlfo7i42PX3n3/+WbNnz9Zf/vIXtWnTRsHBwResCQAkwhoAi0VFRembb75RkyZN9PDDD+uPf/yj\n6tWrpxdeeEGSdMMNNyg8PFx33HGHDh8+rAEDBuitt95SXFycbr/9dv3www+aNGmSpkyZIofDoVmz\nZunQoUOlXmfRokU6cOCAFixY4Hp/2fmOFRYWpjvvvFNdu3bV999/r3/+8586ePCgPv74Y+3du1fz\n58+Xw+HQggULtG/fPi1btkxr1651XQ3ctWuXunbtqlatWqlx48aKiYm5bOcTwJXJYX7LGzEAAOX2\nz3/+U/7+/rrnnnsknX7P2tSpU/XHP/5RoaGhXq4OgK24sgYAl8lLL72kvLw81+P8/HwFBQUR1ABc\nEFfWAOAy+eSTT/TUU08pPz9fjRs3VufOnfX444+rcuXK3i4NgMUIawAAABbjNigAAIDFCGsAAAAW\nI6wBAABYjLAGAABgMcIaAACAxf4fRxUA8AqozSgAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10253f090>"
       ]
      }
     ],
     "prompt_number": 6
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The following function is used to re-compute review counts and averages whenever you subset a reviews data frame. We'll use it soon to construct a smaller, more computationally tractable data frame."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def recompute_frame(ldf):\n",
      "    \"\"\"\n",
      "    takes a dataframe ldf, makes a copy of it, and returns the copy\n",
      "    with all averages and review counts recomputed\n",
      "    this is used when a frame is subsetted.\n",
      "    \"\"\"\n",
      "    ldfu=ldf.groupby('user_id')\n",
      "    ldfb=ldf.groupby('business_id')\n",
      "    user_avg=ldfu.stars.mean()\n",
      "    user_review_count=ldfu.review_id.count()\n",
      "    business_avg=ldfb.stars.mean()\n",
      "    business_review_count=ldfb.review_id.count()\n",
      "    nldf=ldf.copy()\n",
      "    nldf.set_index(['business_id'], inplace=True)\n",
      "    nldf['business_avg']=business_avg\n",
      "    nldf['business_review_count']=business_review_count\n",
      "    nldf.reset_index(inplace=True)\n",
      "    nldf.set_index(['user_id'], inplace=True)\n",
      "    nldf['user_avg']=user_avg\n",
      "    nldf['user_review_count']=user_review_count\n",
      "    nldf.reset_index(inplace=True)\n",
      "    return nldf"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 7
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**1.3** Create a smaller data set in dataframe `smalldf` by looking for those businesses with more than 150 reviews and those users with more than 60 reviews. Include all the columns that were there in the parent dataframe. Since you have created a subset of the data set, use the method provided above to recalculate the averages. Print the number of unique users and items in this data set.  \n",
      "\n",
      "Note that while this cut makes sure we have prolific users, the cut on businesses restores sparsity  by reducing the number of reviews per user."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "smallidf=fulldf[(fulldf.user_review_count > 60) & (fulldf.business_review_count > 150)]\n",
      "smalldf=recompute_frame(smallidf)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 8
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "How does this compare to the parent data set, in terms of size and sparsity? Once again, plot histograms of the review count grouped by user, and by the review count grouped by business, respectively, and describe the results"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "print \"Total Number of Reviews\", smalldf.shape[0]\n",
      "print \"Users in this set\", smalldf.user_id.unique().shape[0], \"Restaurants\",smalldf.business_id.unique().shape[0]\n",
      "plt.figure()\n",
      "ax=smalldf.groupby('user_id').review_id.count().hist()\n",
      "remove_border(ax)\n",
      "plt.xlabel(\"Reviews per user\")\n",
      "plt.grid(False)\n",
      "plt.grid(axis = 'y', color ='white', linestyle='-')\n",
      "plt.figure()\n",
      "ax=smalldf.groupby('business_id').review_id.count().hist()\n",
      "remove_border(ax)\n",
      "plt.xlabel(\"Reviews per restaurant\")\n",
      "plt.grid(False)\n",
      "plt.grid(axis = 'y', color ='white', linestyle='-')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Total Number of Reviews 6165\n",
        "Users in this set 240 Restaurants 172\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAlQAAAGHCAYAAAB22sw1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x107bbe5d0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAlcAAAGHCAYAAACd7Xc2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x107bbebd0>"
       ]
      }
     ],
     "prompt_number": 9
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*your answer here*: As we made a very specific cut on the data set, making sure we had those businesses with greater than 150 reviews, and those users with more than 60 reviews, this data set is not as sparse as the full one, with a range in both graphs where the histogram is somewhat flat rather than steeply declining. "
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**1.4** Compute histograms of the average user rating in the smaller data set, and the average business rating in the smaller data set. Print the overall mean."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "plt.figure()\n",
      "avg_ratings_by_user=smalldf.groupby('user_id').stars.mean()\n",
      "ax=avg_ratings_by_user.hist()\n",
      "remove_border(ax)\n",
      "plt.xlabel(\"Average review score\")\n",
      "plt.grid(False)\n",
      "plt.grid(axis = 'y', color ='white', linestyle='-')\n",
      "plt.title(\"Average User Rating\")\n",
      "plt.figure()\n",
      "\n",
      "avg_ratings_by_biz=smalldf.groupby('business_id').stars.mean()\n",
      "ax=avg_ratings_by_biz.hist()\n",
      "remove_border(ax)\n",
      "plt.xlabel(\"Average review score\")\n",
      "plt.grid(False)\n",
      "plt.grid(axis = 'y', color ='white', linestyle='-')\n",
      "plt.title(\"Average Restaurant Rating\")\n",
      "plt.figure()\n",
      "\n",
      "print smalldf.stars.mean()\n",
      "plt.figure()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "3.86763990268\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 10,
       "text": [
        "<matplotlib.figure.Figure at 0x107bcf150>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAlYAAAGRCAYAAACnoP2OAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x107fb4050>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAlYAAAGRCAYAAACnoP2OAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nWAEAAFiEYgUAAGARihUAAIBF/h+FYNXYvp+UdQAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x107895f50>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "text": [
        "<matplotlib.figure.Figure at 0x107869ad0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "text": [
        "<matplotlib.figure.Figure at 0x107bcf150>"
       ]
      }
     ],
     "prompt_number": 10
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Common Support\n",
      "\n",
      "Lets now make a histogram of the common user support (the number of common reviewers) of each pair of restaurants on the smaller set, and print the mean. Pay attention to the code, as you will use parts of it later. (This code takes a bit of time to run, so be patient).\n",
      "\n",
      "The common support is an important concept, as for each pair of restaurants, its the number of people who reviewed both. It will be used to modify similarity between restaurants. If the common support is low, the similarity is less believable."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "restaurants=smalldf.business_id.unique()\n",
      "supports=[]\n",
      "for i,rest1 in enumerate(restaurants):\n",
      "    for j,rest2 in enumerate(restaurants):\n",
      "        if  i < j:\n",
      "            rest1_reviewers = smalldf[smalldf.business_id==rest1].user_id.unique()\n",
      "            rest2_reviewers = smalldf[smalldf.business_id==rest2].user_id.unique()\n",
      "            common_reviewers = set(rest1_reviewers).intersection(rest2_reviewers)\n",
      "            supports.append(len(common_reviewers))\n",
      "print \"Mean support is:\",np.mean(supports)\n",
      "plt.hist(supports)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Mean support is: 6.84679722562\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 11,
       "text": [
        "(array([  7.02000000e+03,   4.98700000e+03,   1.79400000e+03,\n",
        "         5.90000000e+02,   1.95000000e+02,   7.60000000e+01,\n",
        "         2.20000000e+01,   1.00000000e+01,   1.00000000e+01,\n",
        "         2.00000000e+00]),\n",
        " array([  0. ,   5.1,  10.2,  15.3,  20.4,  25.5,  30.6,  35.7,  40.8,\n",
        "        45.9,  51. ]),\n",
        " <a list of 10 Patch objects>)"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAF1CAYAAABLbYZYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3X9slfX99/HX1XMOUCfn1EmB8uO0RewPt+gGRAsinMYC\nX+U+1VhciBJaXTDSqDFxiUxlg2q2suB3ZMjWrBNZhi5BE9rRLmaw9jSlmJWBTtadHkHC5HQrKrUS\n2gldz3X/4e3Bcwu058A5n9Y+H0kTer2vnvO5rpyUZ65zTo9l27YtAAAApFya6QUAAACMVYQYAACA\nIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIcMKsf379+tHP/qRtmzZolWrVikUCkmSurq6\nVFlZqZqaGpWXl6ujoyP6M4nOAAAAxgprqD/oOjg4qPz8fL333ntKS0tTS0uLXnjhBe3du1dz587V\npk2bVFJSomAwqOXLl+vYsWOyLEvz5s2La3b06FE5HI5UHTcAAIBxzqF26Onp0b/+9S/19/fr2muv\nVUZGhj755BPt27dPwWBQPp9PklRYWCiXy6Xdu3fL7XbHPaurq1NZWVmyjhMAAGDEGfKpyczMTM2d\nO1erV6/WmTNntHXrVj3//PPav3+/cnNz5XReaLm8vDw1NTXpwIEDCc0AAADGkmG9Ruz1119XZ2en\npk2bpjvvvFN33XWXuru75fF4YvbLyMhQOByOe+bxeBQOh6/wUAAAAEaXIZ+alKTu7m6VlJSou7tb\nFRUVcjqdcrlccrlcMftFIhHZth2dxzMDAAAYa4YMsf7+ft111106cuSIJk2apOeee07f//739YMf\n/ECffvppzL69vb3yer3KyspSa2trXLOcnJyv3Pfs2bP1/vvvJ3BYAAAAqXXDDTfo2LFjcf3MkE9N\n/v3vf1ckEtGkSZMkSRs3blRaWpp8Pp+OHz8es29nZ6eKi4tVXFwc1ywUCkVfvP9l77//vmzb5iuF\nXz/+8Y+Nr2GsfXHOOedj4YtzzjkfC1+JXDwaMsRuvPFGnT9/Xv/+978lSefPn9c3vvENfec731F2\ndraam5ujodXX1ye/36+ioqK4Zv39/fL7/XEvHgAAYDQb8qnJ6667Tm+88YaeeuopzZs3TydPntTv\nfvc7ud1u1dfXq6qqSsFgUO3t7WpsbFR6erokxTVraGiIzgAAAMaKIf+gq0mWZWkEL+9rKRAIXPRp\nYiQP5zz1OOepxzlPPc556iXSLYQYAADAVZBIt/Ch3wAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAA\nAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQ\nYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAA\nAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIZcNsROnjwph8OhtLS0mK9QKKSuri5VVlaq\npqZG5eXl6ujoiP5cojMAAICxxLJt277UcNu2bSooKNCNN94oSfrss89UVlamI0eOaO7cudq0aZNK\nSkoUDAa1fPlyHTt2TJZlad68eXHNjh49KofD8dXFWZYuszwAAIARI5FucV5uWFZWpqlTp0a//+Mf\n/6glS5Zo7969CgaD8vl8kqTCwkK5XC7t3r1bbrc77lldXZ3KysriWjgAAMBod9mnJr8cYZJUX1+v\n0tJStbW1KTc3V07nhY7Ly8tTU1OTDhw4kNAMAABgrBn2i/UjkYhaW1t1xx13qLu7Wx6PJ2aekZGh\ncDgc98zj8SgcDl/BIQAAAIxOl31q8sv+8pe/aM6cOXI4HHI6nXK5XDHzSCQi27YTml3O2YFzw13i\niOK00jTB6Rp6RwAAMGYNO8Tq6upUWloqScrKytL+/ftj5r29vfJ6vcrKylJra2tcs5ycnEveb+GD\n/yf6b8+3ZsnzrRuGu2Rj7ph2o56Z9z+aIEIMAICvq0AgoEAgcEW3cdl3TX7ZzTffrLa2Nk2cOFFv\nvfWWli1bpjNnzkTnN9xwg376059q5syZCc2+973vfXVxlqXp25++kuMzYqn3Jm254365x6WbXgoA\nAEiRRN41OazXiAWDQU2ZMkUTJ06UJBUVFSk7O1vNzc2SpM7OTvX19cnv98c96+/vl9/vj2vRAAAA\nXwfDemryD3/4g+65557o95Zlqb6+XlVVVQoGg2pvb1djY6PS0z+/AhTPrKGhIToDAAAYS4b91KQJ\nPDUJAABGi6Q9NQkAAICrjxADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAw\nhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBAD\nAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAwhBADAAAw\nhBADAAAwhBADAAAwhBADAAAwxBnPzidOnNCuXbs0efJkLV++XJmZmclaFwAAwNfesK+I7dq1Sw88\n8IDuv/9+VVRUKDMzU11dXaqsrFRNTY3Ky8vV0dER3T/RGQAAwFgxrCtigUBAjz32mN555x1NmzZN\nkmTbtkpLS7Vp0yaVlJRo8eLFWr58uY4dOybLsuKeHT16VA6HI6kHCwAAMJIMGWK2bWvt2rV64okn\nohEmSfv27VMwGJTP55MkFRYWyuVyaffu3XK73XHP6urqVFZWdtUPEAAAYKQaMsTeeusthUIhnThx\nQitWrFBHR4cee+wxffTRR8rNzZXTeeEm8vLy1NTUpMmTJyc0I8QAAMBYMmSIHTp0SBMnTlR1dbUm\nTZqkw4cP69Zbb9WSJUvk8Xhi9s3IyFA4HFYkEolr5vF4FA6Hr8LhAAAAjB5Dvlj/7Nmzys/P16RJ\nkyRJc+bM0bx58zR79my5XK6YfSORiGzbltPpjHsGAAAw1gx5RWzq1Knq6+uL2TZjxgxt27ZNt9xy\nS8z23t5eeb1eZWVlqbW1Na5ZTk7ORe//TN3+6L/HF3g1vsA71JIBAACSLhAIKBAIXNFtDBli8+fP\n1wcffKCBgYHolaxz585pw4YN2rx5c8y+nZ2dKi8v18yZM1VdXT3sWSgUUkVFxUXv333vwniOBwAA\nICV8Pl/0zYeStHHjxrhvY8inJgsKCjR37lw1NDRIks6fP693331XjzzyiLKzs9Xc3Czp89Dq6+uT\n3+9XUVFRXLP+/n75/f64Fw8AADCaDevviO3cuVNPPfWUQqGQwuGwamtrNXXqVNXX16uqqkrBYFDt\n7e1qbGxUenq6JMU1a2hoiM4AAADGCsu2bdv0Ii7FsixN3/606WXEban3Jm254365xxGXAACMFZZl\nKd6s4kO/AQAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAA\nDCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHE\nAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAADCHEAAAA\nDCHEAAAADCHEAAAADIk7xHp6etTf35+MtQAAAIwpwwqxhQsXKi0tTWlpaVqwYIGuueYadXV1qbKy\nUjU1NSovL1dHR0d0/0RnAAAAY4lzqB0OHTqkZcuW6Re/+IUkacaMGbJtW6Wlpdq0aZNKSkq0ePFi\nLV++XMeOHZNlWXHPjh49KofDkfSDBQAAGEmGDLEtW7bo5ptv1sSJE3XjjTdKkvbu3atgMCifzydJ\nKiwslMvl0u7du+V2u+Oe1dXVqaysLCkHCAAAMFJd9qnJwcFB9fT06MUXX1R+fr5WrlypgYEBtbW1\nadasWXI6L3RcXl6empqadODAAeXm5sY9AwAAGGsue0XM4XCosbFRtm3r1Vdf1dq1a/XMM8/o7Nmz\ncrvdMftmZGQoHA4rEonI4/EMe+bxeBQOh6/S4QAAAIwew3qxvmVZWrVqlX7+859r586dcjqdcrlc\nMftEIhHZtp3QDAAAYCwa8jViX3bPPffo8ccfV1ZWllpbW2Nmvb298nq9Cc1ycnIueZ9n6vZH/z2+\nwKvxBd54lgwAAJAUgUBAgUDgim4jrhAbHBxUfn6+iouLVV1dHTPr7OxUeXm5Zs6cGdcsFAqpoqLi\nkvfpvndhPEscMdId40wv4YoMRAblSuOdrAAAXIrP54u+AVGSNm7cGPdtXDbEDh48qL/97W96+OGH\nlZaWpq1bt+rZZ5/V/PnzlZ2drebmZhUXF6uzs1N9fX3y+/2aMGFCXLP+/n75/f64Fz7SuRwOzXhl\nnellJCz8UPXQOwEAgCty2RDr7u7W+vXrtXPnTi1btky33XabSktLJUn19fWqqqpSMBhUe3u7Ghsb\nlZ6eHvesoaEhOgMAABhLLNu2bdOLuBTLsjR9+9OmlxG3pd6btP3O1VwRAwBgDLEsS/FmFR/6DQAA\nYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAgh\nBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAA\nYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAghBgAAYAgh\nBgAAYAghBgAAYMiwQywSiai4uFgtLS2SpK6uLlVWVqqmpkbl5eXq6OiI7pvoDAAAYCxxDnfHX/3q\nV3r33XdlWZZs21Zpaak2bdqkkpISLV68WMuXL9exY8dkWVbcs6NHj8rhcCTzOAEAAEacYYXY/v37\nlZubK7fbLUnat2+fgsGgfD6fJKmwsFAul0u7d++W2+2Oe1ZXV6eysrKrfnAAAAAj2ZBPTZ4+fVoH\nDhzQ3XffLUmybVttbW3Kzc2V03mh4/Ly8tTU1KQDBw4kNAMAABhrhrwitmXLFq1fvz5m26lTp+Tx\neGK2ZWRkKBwOKxKJxDXzeDwKh8OJrh8AAGDUuuwVsdraWj344IMaN25czHaHwyGXyxWzLRKJyLZt\nOZ3OuGcAAABj0WWviNXW1uqJJ56Ifn/u3DktXbpUtm3rW9/6Vsy+vb298nq9ysrKUmtra1yznJyc\nS67hTN3+6L/HF3g1vsA75EEBAAAkWyAQUCAQuKLbuGyItbe3x3yfm5ur3/72t3K5XFq2bFnMrLOz\nU+Xl5Zo5c6aqq6uHPQuFQqqoqLjkGtz3LhzOcQAAAKSUz+eLvgFRkjZu3Bj3bST0B12LioqUnZ2t\n5uZmSZ+HVl9fn/x+f9yz/v5++f3+RJYBAAAwqg3774h9mWVZqq+vV1VVlYLBoNrb29XY2Kj09HRJ\nimvW0NAQnQEAAIwllm3btulFXIplWZq+/WnTy4jbUu9N2n7nas14ZZ3ppSQs/FD10DsBAICoL/7o\nfTz4rEkAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABD\nCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEA\nAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABDCDEAAABD\nCDEAAABDCDEAAABDCDEAAABDhhVib7/9tm6//XZdd911WrJkiU6fPi1J6urqUmVlpWpqalReXq6O\njo7ozyQ6AwAAGCuGDLHz58/r9ddf1759+xQOh3X27Fn97//+rySptLRU9913nx599FGtW7dOfr9f\nkUhEtm3HPRscHEz6wQIAAIwkzqF2+OSTT7RhwwaNGzdOkrR48WI5HA7t3btXwWBQPp9PklRYWCiX\ny6Xdu3fL7XbHPaurq1NZWVlSDhIAAGAkGvKK2JQpU6IRdu7cOZ06dUpPPvmk2traNGvWLDmdF1ou\nLy9PTU1NOnDggHJzc+OeAQAAjCVDXhH7wp49e/Tcc8+pp6dHHR0d6u7ultvtjtknIyND4XBYkUhE\nHo9n2DOPx6NwOHwFhwEAADD6DPtdk36/X/X19Vq0aJFWrVoll8sll8sVs88XrwFzOp1xzwAAAMaa\nYV8Rk6ScnBy9/PLLuv7665WZmalPP/00Zt7b2yuv16usrCy1trbGNcvJybnofZ6p2x/99/gCr8YX\neONZMgAAQFIEAgEFAoEruo24QkySJkyYoOuvv14lJSXavHlzzKyzs1Pl5eWaOXOmqqurhz0LhUKq\nqKi46P25710Y7xIBAACSzufzRd98KEkbN26M+zaGfGqyp6dHe/bsiX7f0tKi1atXa8GCBcrOzlZz\nc7Okz0Orr69Pfr9fRUVFcc36+/vl9/vjXjwAAMBoNuQVsePHj2vNmjXKz8/XihUrdO211+qFF16Q\nJNXX16uqqkrBYFDt7e1qbGxUenp63LOGhoboDAAAYKywbNu2TS/iUizL0vTtT5teRtyWem/S9jtX\na8Yr60wvJWHhh6qH3gkAAERZlqV4s4rPmgQAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMA\nADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCE\nEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMA\nADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADCEEAMAADBkyBBraWnRLbfcIrfbrWXLlunk\nyZOSpK6uLlVWVqqmpkbl5eXq6OiI/kyiMwAAgLHksiH24Ycfavv27Xr11Vf1+uuvKxQK6eGHH5Yk\nlZaW6r777tOjjz6qdevWye/3KxKJyLbtuGeDg4MpOVgAAICRxHm5YVNTk1566SVNnDhR3/72t7Vh\nwwatXbtW+/btUzAYlM/nkyQVFhbK5XJp9+7dcrvdcc/q6upUVlaWzOMEAAAYcS4bYitXroz5fsqU\nKfJ6vWpra1Nubq6czgs/npeXp6amJk2ePDmhGSEGAADGmsuG2P/v8OHDWrt2rUKhkDweT8wsIyND\n4XBYkUgkrpnH41E4HE5w+QAAAKPXsN812dfXpyNHjujxxx+Xw+GQy+WKmX/xGjCn0xn3DAAAYCwa\n9hWxzZs3a+vWrXI4HJo2bZr2798fM+/t7ZXX61VWVpZaW1vjmuXk5Fzyfs/UXbif8QVejS/wDnfJ\nAAAASRMIBBQIBK7oNoYVYrW1tVq1apUyMzMlSQsXLlR1dXXMPp2dnSovL9fMmTPjmoVCIVVUVFzy\nvt33LhzOEgEAAFLK5/NF34AoSRs3boz7NoZ8anLHjh1KT0/XwMCAOjs71dLSouPHjysnJ0fNzc2S\nPg+tvr4++f1+FRUVKTs7e9iz/v5++f3+uBcOAAAw2l32itibb76pNWvWxPydL8uyFAqFtGjRIlVV\nVSkYDKq9vV2NjY1KT0+XJNXX1w971tDQEJ0BAACMJZZt27bpRVyKZVmavv1p08uI21LvTdp+52rN\neGWd6aUkLPxQ9dA7AQCAKMuyFG9W8VmTAAAAhhBiAAAAhhBiAAAAhhBiuKiByOj9IPbRvHYAwNgS\n10ccYexwpTlG7ZsNeKMBAGC04IoYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQY\nAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACA\nIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIYQYAACAIXGF\n2GeffaYzZ84kay0AAABjyrBCzLZt7dixQ3l5eTp48GB0e1dXlyorK1VTU6Py8nJ1dHRc8QwAAGCs\nGFaIffzxxyopKVE4HJZlWZI+j7PS0lLdd999evTRR7Vu3Tr5/X5FIpGEZoODg0k9UAAAgJHGOZyd\nMjMzv7Jt3759CgaD8vl8kqTCwkK5XC7t3r1bbrc77lldXZ3KysquykEBAACMBgm/WL+trU2zZs2S\n03mh5fLy8tTU1KQDBw4oNzc37hkAAMBYMqwrYhfT3d0tt9sdsy0jI0PhcFiRSEQej2fYM4/Ho3A4\nnOhSAAAARqWEr4g5nU65XK6YbV+8BiyRGQAAwFiT8BWxadOmaf/+/THbent75fV6lZWVpdbW1rhm\nOTk5F72fM3UX7mN8gVfjC7yJLhkAAOCqCQQCCgQCV3QbCYeYz+dTdXV1zLbOzk6Vl5dr5syZcc1C\noZAqKiouej/uexcmukQAAICk8fl80TcfStLGjRvjvo1hPzX5xdOHtm1LkubPn6/s7Gw1NzdL+jy0\n+vr65Pf7VVRUFNesv79ffr8/7sUDAACMZsO6IvbRRx+ptrZWlmXptdde0/Tp01VQUKD6+npVVVUp\nGAyqvb1djY2NSk9Pl6S4Zg0NDdEZAADAWGHZX1ziGoEsy9L07U+bXkbclnpv0vY7V2vGK+tMLyVh\n4YeqR+36ww9VD70TAABXmWVZijer+NBvAAAAQwgxAAAAQwgxAAAAQwgxAAAAQwgxAAAAQwgxAAAA\nQwgxAAAAQwgxAAAAQwgxAAAAQwgxAAAAQwgxAAAAQwgxAAAAQwgxfO0MRAZNLyFho3ntAID4OU0v\nALjaXGkOzXhlnellJCT8ULXpJQAAUogrYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAA\nAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYgAAAIYQYsAIMhAZ\nNL2EKzLa1w8AqeY0vQAAF7jSHJrxyjrTy0hY+KFq00sAgFHFyBWxrq4uVVZWqqamRuXl5ero6DCx\nDAAAAKNSfkXMtm2VlpZq06ZNKikp0eLFi7V8+XIdPXpUDocj1csBAAAwJuVXxPbt26dgMCifzydJ\nKiwslMvlUl1dXaqXgos41/mB6SWMOZzz1AsEAqaXMOZwzlOPcz46pDzE2traNGvWLDmdFy7G5eXl\nqampKdVLwUUQBan3dTrno+XF+hf7D2q0rH20IgpSj3M+OqT8qcnu7m653e6YbR6PR+FwONVLAXCV\njZY3G5x5e79+88pnMdt4owEAE1IeYk6nUy6XK2ZbJBK55P4Z469J9pKuuomu8aaXAAAARgHLtm07\nlXf4k5/8RLt27dI777wT3Xb33XcrJydHv/zlL2P2nT17tt5///1ULg8AACAhN9xwg44dOxbXz6T8\nilhxcbGqq2OfAgiFQqqoqPjKvvEeDAAAwGiS8hfrFxUVKTs7W83NzZKkzs5O9ff3y+/3p3opAAAA\nRqX8iphlWaqvr1dVVZWCwaDa29vV0NCg9PT0VC8FAADAqJS/Rgwjy2effabz589/5Z2sAIDRqaen\nRxMmTNA114y+N7uNNidOnNCuXbs0efJkLV++XJmZmXHfxoj80G8+Ain5bNvWjh07lJeXp4MHD0a3\nc+6Tp6WlRbfccovcbreWLVumkydPSuKcJ9Pbb7+t22+/Xdddd52WLFmi06dPS+Kcp0IkElFxcbFa\nWlokcc6TbeHChUpLS1NaWpoWLFiga665hnOeZLt27dIDDzyg+++/XxUVFcrMzEzsnNsjTCQSsefM\nmWPv3bvXtm3b/sc//mHn5uba//3vfw2v7Ovlww8/tE+ePGlblmX/+c9/tm2bc59Mp06dslevXm0f\nOXLEfvPNN+3s7Gy7pKTEtm2bc54k586ds3/4wx/a/f399tmzZ+2ioiL7mWeesW2bc54KL730kv3N\nb37Tbmlp4XdLkv31r3+1q6qq7EOHDtmHDh2yT506xTlPsubmZjszM9Pu6uqKbkv0nI+4EPvTn/5k\np6en2wMDA9FteXl59htvvGFwVV9fXw4xzn3y/P73v7fPnDkT/f6VV16xJ0yYYO/du5dzniTd3d32\nuXPnot8//fTT9vr163mcp0Bra6vd2Nho5+Tk2C0tLZzzJFu1apX9s5/9zH7vvfei2zjnyROJROyC\nggL7+eefj9me6DkfcU9N8hFI5nDuk2flypWaOHFi9PspU6bI6/Wqra1Nubm5nPMkmDJlisaNGydJ\nOnfunE6dOqUnn3ySx3mSnT59WgcOHNDdd98t6fOXQfA4T57BwUH19PToxRdfVH5+vlauXKmBgQEe\n50n01ltvKRQK6cSJE1qxYoUKCwu1bdu2hB/nKX/X5FD4CCRzOPepc/jwYa1du1ahUEgejydmxjm/\nuvbs2aPnnntOPT096ujo4HGeZFu2bNH69etjtp06dYrHeZI4HA41NjbKtm29+uqrWrt2rZ555hmd\nPXuWx3mSHDp0SBMnTlR1dbUmTZqkw4cP69Zbb9WSJUsSepyPuCti8X4EEq4ezn1q9PX16ciRI3r8\n8cflcDg450nm9/tVX1+vRYsWadWqVXK5XJzzJKmtrdWDDz4YvRL5BR7nyWdZllatWqWf//zn2rlz\nJ7/Pk+js2bPKz8/XpEmTJElz5szRvHnzNHv27ITO+YgLsWnTpunTTz+N2dbb26vp06cbWtHYwblP\njc2bN2vr1q1yOByc8xTJycnRyy+/rI8//liZmZmc8ySpra3Vd7/7XaWnpys9PV3//Oc/tXTpUv36\n17/WmTNnYvblnCfHPffco97eXmVlZfE4T5KpU6eqr68vZtuMGTO0bdu2hB7nIy7EiouLdfz48Zht\noVBIPp/PzILGEJ/Px7lPstraWq1atSr6t2YWLlzIOU+RCRMm6Prrr1dJSQnnPEna29v1n//8J/qV\nnZ2tvXv3qqWl5SufG8w5T47BwUHl5+fzf2kSzZ8/Xx988IEGBgai286dO6cNGzYk9DgfcSHGRyCl\nzheXTO3jIrIyAAABQ0lEQVT/9zd958+fz7lPoh07dig9PV0DAwPq7OxUS0uLjh8/rpycHM55EvT0\n9GjPnj3R71taWrR69WotWLCAx3mK8Xs9eQ4ePKjf/OY30d/nW7du1bPPPsvv8yQqKCjQ3Llz1dDQ\nIEk6f/683n33XT3yyCMJnfMR92J9PgIpNT766CPV1tbKsiy99tprmj59ugoKCjj3SfLmm29qzZo1\nGhwcjG6zLEuhUEiLFi3inCfB8ePHtWbNGuXn52vFihW69tpr9cILL0gSj/MU4/d68nR3d2v9+vXa\nuXOnli1bpttuu02lpaWSeJwn086dO/XUU08pFAopHA6rtrZWU6dOTeic8xFHAAAAhoy4pyYBAADG\nCkIMAADAEEIMAADAEEIMAADAEEIMAADAEEIMAADAEEIMAADAEEIMAADAEEIMAADAkP8LYP/402N5\nDYAAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x107be99d0>"
       ]
      }
     ],
     "prompt_number": 11
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "As you can see, even though we chose a subset of the dataframe in which every restaurant had 150 reviews and every user had atleast made 60, the common support of most pairs of restaurants is really low, indeed less than 10!."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "### Calculating Similarity"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Users rate restaurants on a scale of 1-5. Even though this rating is integer valued, for the purposes of this assignment we shall treat it as a real number.\n",
      "\n",
      "Even though each reviewer uses the same 5-star scale when rating restaurants, comparing two users by comparing their raw user ratings can be problematic. Consider a user whose average rating is 2. This is a curmudgeonly user. Consider another whose average rating is 4. This is a rather enthusiastic one. How should we compare a 3 rating by the curmudgeonly one to a 5 rating of the enthusiastic one?\n",
      "\n",
      "It is for this purpose that we must subtract the average rating of the user from the actual rating of the restaurants in computing the similarity of two restaurants. This makes the above ratings by the two users comparable. We do this in the function `pearson_sim` defined below.\n",
      "\n",
      "If there is no common support (`n_common=0`), we have no basis for making a similarity estimate, and so we set the similarity to 0. In the case that the individual restaurant rating variance is 0, such as in the case where there is only one common reviewer (`n_common=1`), we return the `NaN` that the scipy `pearsonr` returns. We will deal with it soon,"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from scipy.stats.stats import pearsonr\n",
      "def pearson_sim(rest1_reviews, rest2_reviews, n_common):\n",
      "    \"\"\"\n",
      "    Given a subframe of restaurant 1 reviews and a subframe of restaurant 2 reviews,\n",
      "    where the reviewers are those who have reviewed both restaurants, return \n",
      "    the pearson correlation coefficient between the user average subtracted ratings.\n",
      "    The case for zero common reviewers is handled separately. Its\n",
      "    ok to return a NaN if any of the individual variances are 0.\n",
      "    \"\"\"\n",
      "    if n_common==0:\n",
      "        rho=0.\n",
      "    else:\n",
      "        diff1=rest1_reviews['stars']-rest1_reviews['user_avg']\n",
      "        diff2=rest2_reviews['stars']-rest2_reviews['user_avg']\n",
      "        rho=pearsonr(diff1, diff2)[0]\n",
      "    return rho"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 12
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The function `get_restaurant_reviews` defined below takes a restaurant `business_id` and a set of users, and returns the reviews of that restaurant by those users. You will use this function in calculating a similarity function, in **1.5**."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def get_restaurant_reviews(restaurant_id, df, set_of_users):\n",
      "    \"\"\"\n",
      "    given a resturant id and a set of reviewers, return the sub-dataframe of their\n",
      "    reviews.\n",
      "    \"\"\"\n",
      "    mask = (df.user_id.isin(set_of_users)) & (df.business_id==restaurant_id)\n",
      "    reviews = df[mask]\n",
      "    reviews = reviews[reviews.user_id.duplicated()==False]\n",
      "    return reviews"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 13
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**1.5** Write a function `calculate_similarity` that operates between two restaurants and calculates a similarity for them, taking a dataframe and a similarity function `similarity_func`. An example of the `similarity_func` is the `pearson_sim` we defined above. `calculate_similarity` operates as follows: \n",
      "\n",
      "1. For each of the two restaurants, get the set of reviewers who have reviewed the restaurant and compute the intersection of these two sets. Also compute the number of common reviewers `n_common`.\n",
      "\n",
      "2. Use the function `get_restaurant_reviews` defined below to get the reviews for each restaurant as made by these common reviewers. Notice that `get_restaurant_reviews` returns a sub data frame of reviews.\n",
      "\n",
      "3. Calculate the similarity using `similarity_func` which takes the two reviews dataframes from part 2 and the number of common reviewers `n_common` as arguments\n",
      "\n",
      "4. Return the similarity and `n_common` in a tuple `(sim, n_common)`. If the similarity is a `NaN`, set the similarity to 0.\n"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "\"\"\"\n",
      "Function\n",
      "--------\n",
      "calculate_similarity\n",
      "\n",
      "Parameters\n",
      "----------\n",
      "rest1 : string\n",
      "    The id of restaurant 1\n",
      "rest2 : string\n",
      "    The id of restaurant 2\n",
      "df : DataFrame\n",
      "  A dataframe of reviews, such as the smalldf above\n",
      "similarity_func : func\n",
      "  A function like pearson_sim above which takes two dataframes of individual\n",
      "  restaurant reviews made by a common set of reviewers, and the number of\n",
      "  common reviews. This function returns the similarity of the two restaurants\n",
      "  based on the common reviews.\n",
      "  \n",
      "Returns\n",
      "--------\n",
      "A tuple\n",
      "  The first element of the tuple is the similarity and the second the\n",
      "  common support n_common. If the similarity is a NaN, set it to 0\n",
      "\"\"\"\n",
      "#your code here\n",
      "def calculate_similarity(rest1, rest2, df, similarity_func):\n",
      "    # find common reviewers\n",
      "    rest1_reviewers = df[df.business_id==rest1].user_id.unique()\n",
      "    rest2_reviewers = df[df.business_id==rest2].user_id.unique()\n",
      "    common_reviewers = set(rest1_reviewers).intersection(rest2_reviewers)\n",
      "    n_common=len(common_reviewers)\n",
      "    #get reviews\n",
      "    rest1_reviews = get_restaurant_reviews(rest1, df, common_reviewers)\n",
      "    rest2_reviews = get_restaurant_reviews(rest2, df, common_reviewers)\n",
      "    sim=similarity_func(rest1_reviews, rest2_reviews, n_common)\n",
      "    if np.isnan(sim):\n",
      "        return 0, n_common\n",
      "    return sim, n_common"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 14
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "### Making a database of similarities\n",
      "\n",
      "We now move to calculating a global database of pairwise restaurant similarities.\n",
      "We provide you here with a function to make a database of the similarities for each pair of restaurants in the database. The class `Database` is initialized in its constructor by taking as arguments a dataframe of reviews. The method `populate_by calculating` iterates over every possible pair of business_id's in the dataframe and populates the database with similarities and common supports. It takes as arguments a function the similarity function `similarity_func` like `pearson_sim` (`calculate_similarity` then uses this to calculate the similarity). The `get` method on the database can be used to retrieve the similarity for two business ids."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "(See Thu Oct 17th's class video for information about classes)"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "class Database:\n",
      "    \"A class representing a database of similaries and common supports\"\n",
      "    \n",
      "    def __init__(self, df):\n",
      "        \"the constructor, takes a reviews dataframe like smalldf as its argument\"\n",
      "        database={}\n",
      "        self.df=df\n",
      "        self.uniquebizids={v:k for (k,v) in enumerate(df.business_id.unique())}\n",
      "        keys=self.uniquebizids.keys()\n",
      "        l_keys=len(keys)\n",
      "        self.database_sim=np.zeros([l_keys,l_keys])\n",
      "        self.database_sup=np.zeros([l_keys, l_keys], dtype=np.int)\n",
      "        \n",
      "    def populate_by_calculating(self, similarity_func):\n",
      "        \"\"\"\n",
      "        a populator for every pair of businesses in df. takes similarity_func like\n",
      "        pearson_sim as argument\n",
      "        \"\"\"\n",
      "        items=self.uniquebizids.items()\n",
      "        for b1, i1 in items:\n",
      "            for b2, i2 in items:\n",
      "                if i1 < i2:\n",
      "                    sim, nsup=calculate_similarity(b1, b2, self.df, similarity_func)\n",
      "                    self.database_sim[i1][i2]=sim\n",
      "                    self.database_sim[i2][i1]=sim\n",
      "                    self.database_sup[i1][i2]=nsup\n",
      "                    self.database_sup[i2][i1]=nsup\n",
      "                elif i1==i2:\n",
      "                    nsup=self.df[self.df.business_id==b1].user_id.count()\n",
      "                    self.database_sim[i1][i1]=1.\n",
      "                    self.database_sup[i1][i1]=nsup\n",
      "                    \n",
      "\n",
      "    def get(self, b1, b2):\n",
      "        \"returns a tuple of similarity,common_support given two business ids\"\n",
      "        sim=self.database_sim[self.uniquebizids[b1]][self.uniquebizids[b2]]\n",
      "        nsup=self.database_sup[self.uniquebizids[b1]][self.uniquebizids[b2]]\n",
      "        return (sim, nsup)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 15
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Lets run `make_database` and store the result in the global variable `db`. Lets print out an example entry. Running this function will take a bit of time."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "db=Database(smalldf)\n",
      "db.populate_by_calculating(pearson_sim)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 16
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "db.get(\"z3yFuLVrmH-3RJruPEMYKw\", \"zruUQvFySeXyEd7_rQixBg\")"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 17,
       "text": [
        "(0.39904554525734559, 7)"
       ]
      }
     ],
     "prompt_number": 17
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "### K-Nearest restaurants (in similarity)"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "We are now going to find the k-nearest restaurants to a given restaurant based on the database of similarities that we calculated. But we have a problem.\n",
      "\n",
      "Consider the two cases where there is just one common reviewer, and where there are 40. In the former case, we might get a artificially high similarity based on the tastes of just this user, and thus we must reduce its importance in the nearest-neighbor calculation. In the latter case, we would get a much more unbiased estimator of the similarity of the two restaurants.\n",
      "\n",
      "To control the effect of small common supports, we can **shrink** our pearson co-efficients. We shall do this by using the \"regularization\" parameter `reg`:\n",
      "\n",
      "$$s_{mj} = \\frac{N_{common}\\, \\rho_{mj}}{N_{common}+reg} $$\n",
      "\n",
      "where $N_{common}$ (`n_common`) is the common reviewer support and $\\rho_{ij}$ is the pearson co-relation coefficient.\n",
      "\n",
      "Recall the notions of regularization introduced in class. We want to reduce the variance in our estimates, so we pull our estimates in toward a conservative point in a way that strongly corrals in estimates when there is very little data, but allows the data to speak when there is a lot. This can be shown as equivalent to adding in a `reg` amount of bayesian prior, as Joe has alluded to in class. \n",
      "\n",
      "A good value of the regularizer is intuitively one that dosent affect the similarity when the common support is high ~ 10, but has a large effect when the support is small. In this case, values of 2-4 are good. Usually, the value of `reg` is determined using cross-validation, but for the sake of simplicity we will generally set it to 3.\n",
      "\n",
      "We define a function `shrunk_sim` which takes the `sim` and `n_common` obtained from the database, and shrinks the similarity down using the regularizer `reg`."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def shrunk_sim(sim, n_common, reg=3.):\n",
      "    \"takes a similarity and shrinks it down by using the regularizer\"\n",
      "    ssim=(n_common*sim)/(n_common+reg)\n",
      "    return ssim"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 18
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**1.6** Now we can move to writing a `knearest` function, which finds the `k` nearest neighbors of a given restaurant based on the shrunk similarities we calculate. Note that as defined here, the nearest neighbors are global over the entire set of restaurants, as opposed to being restricted to the restaurants a user has reviewed(we shall do that in the next problem). Thus, this is an expensive function!\n",
      "\n",
      "Write a `knearest` that returns a *k-length sorted list* of 3-tuples each corresponding to a restaurant. The tuple structure is `(business_id, shrunken similarity score, common support)` where the similarity score and common support are with respect to the restaurant whose neighbors we are finding, and the `business_id` is the id of the \"nearby\" restaurant found. The nearby restaurants are found from a supplied numpy array of restaurants `set_of_restaurants`. The spec for the function is given below. HINT: use `itemgetter` from the `operator` module to do the sorting."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "\"\"\"\n",
      "Function\n",
      "--------\n",
      "knearest\n",
      "\n",
      "Parameters\n",
      "----------\n",
      "restaurant_id : string\n",
      "    The id of the restaurant whose nearest neighbors we want\n",
      "set_of_restaurants : array\n",
      "    The set of restaurants from which we want to find the nearest neighbors\n",
      "dbase : instance of Database class.\n",
      "    A database of similarities, on which the get method can be used to get the similarity\n",
      "  of two businessed. e.g. dbase.get(rid1,rid2)\n",
      "k : int\n",
      "    the number of nearest neighbors desired, default 7\n",
      "reg: float\n",
      "    the regularization.\n",
      "    \n",
      "  \n",
      "Returns\n",
      "--------\n",
      "A sorted list\n",
      "    of the top k similar restaurants. The list is a list of tuples\n",
      "    (business_id, shrunken similarity, common support).\n",
      "\"\"\"\n",
      "#your code here\n",
      "from operator import itemgetter\n",
      "def knearest(restaurant_id, set_of_restaurants, dbase, k=7, reg=3.):\n",
      "    \"\"\"\n",
      "    Given a restaurant_id, dataframe, and database, get a sorted list of the\n",
      "    k most similar restaurants from the entire database.\n",
      "    \"\"\"\n",
      "    similars=[]\n",
      "    for other_rest_id in set_of_restaurants:\n",
      "        if other_rest_id!=restaurant_id:\n",
      "            sim, nc=dbase.get(restaurant_id, other_rest_id)\n",
      "            ssim=shrunk_sim(sim, nc, reg=reg)\n",
      "            similars.append((other_rest_id, ssim, nc ))\n",
      "    similars=sorted(similars, key=itemgetter(1), reverse=True)\n",
      "    return similars[0:k]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 19
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "### Ok it's time to recommend!\n",
      "\n",
      "Lets choose the two very different businesses in the dataframe"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "testbizid=\"eIxSLxzIlfExI6vgAbn2JA\"\n",
      "testbizid2=\"L-uPZxooP_ziXCtRrWi8Pw\""
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 20
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "We provide functions to look up a business name given a business id, and a username given a user id."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def biznamefromid(df, theid):\n",
      "    return df['biz_name'][df['business_id']==theid].values[0]\n",
      "def usernamefromid(df, theid):\n",
      "    return df['user_name'][df['user_id']==theid].values[0]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 21
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print testbizid, biznamefromid(smalldf,testbizid)\n",
      "print testbizid2, biznamefromid(smalldf, testbizid2)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "eIxSLxzIlfExI6vgAbn2JA Lobbys Beef Burgers Dogs\n",
        "L-uPZxooP_ziXCtRrWi8Pw Caf\u00e9 Monarch\n"
       ]
      }
     ],
     "prompt_number": 22
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "####Get top matches\n",
      "\n",
      "Its now time to answer the question: \"if you liked this, you might also like these\". We use our `testbizid` and `testbizid2` to compute the `k=7` nearest neighbors with a regularization of `3.` . We print these top 7 matches names, along with their similarity coefficient and common support."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "tops=knearest(testbizid, smalldf.business_id.unique(), db, k=7, reg=3.)\n",
      "print \"For \",biznamefromid(smalldf, testbizid), \", top matches are:\"\n",
      "for i, (biz_id, sim, nc) in enumerate(tops):\n",
      "    print i,biznamefromid(smalldf,biz_id), \"| Sim\", sim, \"| Support\",nc"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "For  Lobbys Beef Burgers Dogs , top matches are:\n",
        "0 La Condesa Gourmet Taco Shop | Sim 0.598714448434 | Support 6\n",
        "1 Citizen Public House | Sim 0.571428571429 | Support 4\n",
        "2 FnB | Sim 0.527129890943 | Support 5\n",
        "3 Defalco's Italian Grocery | Sim 0.519456555658 | Support 6\n",
        "4 Republic Ramen + Noodles | Sim 0.519140146937 | Support 5\n",
        "5 unPhogettable | Sim 0.5 | Support 3\n",
        "6 Haus Murphy's | Sim 0.467637235308 | Support 3\n"
       ]
      }
     ],
     "prompt_number": 23
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "tops2=knearest(testbizid2, smalldf.business_id.unique(), db, k=7, reg=3.)\n",
      "print \"For \",biznamefromid(smalldf, testbizid2), \", top matches are:\"\n",
      "for i, (biz_id, sim, nc) in enumerate(tops2):\n",
      "    print i,biznamefromid(smalldf,biz_id), \"| Sim\", sim, \"| Support\",nc"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "For  Caf\u00e9 Monarch , top matches are:\n",
        "0 Postino Arcadia | Sim 0.625 | Support 5\n",
        "1 The Main Ingredient Ale House And Caf\u00e9 | Sim 0.571428571429 | Support 4\n",
        "2 Brio Tuscan Grille | Sim 0.571428571429 | Support 4\n",
        "3 Kazimierz World Wine Bar | Sim 0.5 | Support 3\n",
        "4 Harlow's Cafe | Sim 0.5 | Support 3\n",
        "5 The Fry Bread House | Sim 0.5 | Support 3\n",
        "6 Cien Agaves Tacos & Tequila | Sim 0.5 | Support 3\n"
       ]
      }
     ],
     "prompt_number": 24
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "We can see that these two restaurants are in somewhat different orbits :-)."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Lets now turn our attention to another question: what are the top recommendations for a user? To answer this we must find the user's top rated restaurants, find the nearest neighbors of these restaurants, merge these lists while removing the duplicates and the ones that the user has already rated, and sort by the restaurant's average rating. We provide the code to get the user's top choices in a subset data frame."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def get_user_top_choices(user_id, df, numchoices=5):\n",
      "    \"get the sorted top 5 restaurants for a user by the star rating the user gave them\"\n",
      "    udf=df[df.user_id==user_id][['business_id','stars']].sort(['stars'], ascending=False).head(numchoices)\n",
      "    return udf\n",
      "testuserid=\"7cR92zkDv4W3kqzii6axvg\"\n",
      "print \"For user\", usernamefromid(smalldf,testuserid), \"top choices are:\" \n",
      "bizs=get_user_top_choices(testuserid, smalldf)['business_id'].values\n",
      "[biznamefromid(smalldf, biz_id) for biz_id in bizs]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "For user Vern top choices are:\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 25,
       "text": [
        "['Local Breeze',\n",
        " \"Carly's Bistro\",\n",
        " 'Tee Pee Mexican Food',\n",
        " 'District American Kitchen and Wine Bar',\n",
        " 'Los Reyes de la Torta']"
       ]
      }
     ],
     "prompt_number": 25
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "#### Get top recommendations for user."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**1.7** Its your job now to write a function `get_top_recos_for_user` which takes as arguments a `userid`, the `n` top choices for the user, the dataframe, `k`, and a regularizer, and returns the top recommendations obtained from combining the restaurants that are neighbors of each of the `n` choices, in the way described in the previous paragraph. This returned list is a list of tuples `(restaurant_id, business_avg)` sorted by `business_avg` where `business_avg` is the average rating of the restaurant over the dataframe."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "\"\"\"\n",
      "Function\n",
      "--------\n",
      "get_top_recos_for_user\n",
      "\n",
      "Parameters\n",
      "----------\n",
      "userid : string\n",
      "    The id of the user for whom we want the top recommendations\n",
      "df : Dataframe\n",
      "    The dataframe of restaurant reviews such as smalldf\n",
      "dbase : instance of Database class.\n",
      "    A database of similarities, on which the get method can be used to get the similarity\n",
      "  of two businesses. e.g. dbase.get(rid1,rid2)\n",
      "n: int\n",
      "    the n top choices of the user by star rating\n",
      "k : int\n",
      "    the number of nearest neighbors desired, default 8\n",
      "reg: float\n",
      "    the regularization.\n",
      "    \n",
      "  \n",
      "Returns\n",
      "--------\n",
      "A sorted list\n",
      "    of the top recommendations. The list is a list of tuples\n",
      "    (business_id, business_avg). You are combining the k-nearest recommendations \n",
      "    for each of the user's n top choices, removing duplicates and the ones the user\n",
      "    has already rated.\n",
      "\"\"\"\n",
      "#your code here\n",
      "def get_top_recos_for_user(userid, df, dbase, n=5, k=7, reg=3.):\n",
      "    bizs=get_user_top_choices(userid, df, numchoices=n)['business_id'].values\n",
      "    rated_by_user=df[df.user_id==userid].business_id.values\n",
      "    tops=[]\n",
      "    for ele in bizs:\n",
      "        t=knearest(ele, df.business_id.unique(), dbase, k=k, reg=reg)\n",
      "        for e in t:\n",
      "            if e[0] not in rated_by_user:\n",
      "                tops.append(e)\n",
      "\n",
      "    #there might be repeats. unique it\n",
      "    ids=[e[0] for e in tops]\n",
      "    uids={k:0 for k in list(set(ids))}\n",
      "\n",
      "    topsu=[]\n",
      "    for e in tops:\n",
      "        if uids[e[0]] == 0:\n",
      "            topsu.append(e)\n",
      "            uids[e[0]] =1\n",
      "    topsr=[]     \n",
      "    for r, s,nc in topsu:\n",
      "        avg_rate=df[df.business_id==r].stars.mean()\n",
      "        topsr.append((r,avg_rate))\n",
      "        \n",
      "    topsr=sorted(topsr, key=itemgetter(1), reverse=True)\n",
      "\n",
      "    if n < len(topsr):\n",
      "        return topsr[0:n]\n",
      "    else:\n",
      "        return topsr"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 26
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Lets print the top recommendations for `testuserid`, with a regularization of 3."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print \"For user\", usernamefromid(smalldf,testuserid), \"the top recommendations are:\"\n",
      "toprecos=get_top_recos_for_user(testuserid, smalldf, db, n=5, k=7, reg=3.)\n",
      "for biz_id, biz_avg in toprecos:\n",
      "    print biznamefromid(smalldf,biz_id), \"| Average Rating |\", biz_avg"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "For user Vern the top recommendations are:\n",
        "Rokerij | Average Rating | 4.37931034483\n",
        "Wildfish Seafood Grille | Average Rating | 4.29411764706\n",
        "Cornish Pasty Company | Average Rating | 4.20689655172\n",
        "Pappadeaux Seafood Kitchen | Average Rating | 4.18518518519\n",
        "Four Peaks Brewing Co | Average Rating | 4.16666666667\n"
       ]
      }
     ],
     "prompt_number": 27
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "## Problem 2: A user based recommender with predicted ratings"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "This is all very nice. We can provide ratings based on global similarities to a restaurant. However, in many cases this is not enough.\n",
      "\n",
      "For example, it is hard to judge if the above recommendations are any good. In the usual testing paradigm, say that we break the dataframe into train and test. Based on the training set, I am recommended restaurant B. Now, I have rated B, but that information is in the testing set. I have no way of comparing the rating I give B in the testing set, to the similarity computed from the training set that was used to make the recomendation. The best I could do is to compare the average rating of restaurant B in the training set to my rating of restaurant B in the test set. \n",
      "\n",
      "In this section, we shift our focus to more fine-grained predictions about each user, and try to predict what _rating_ a user would give to a restaurant they have never tried before. To do this, we will try to personalize the information we use even further, and only pool information from restaurants that the user has rated.\n",
      "\n",
      "This allows us to return to the original problem of prediction $Y_{um}$ for a restaurant $m$ that user $u$ has never rated before. Using our newly computed similarity metrics, we can modify our original baseline estimate by pulling in information from the user's neighborhood of the restaurant $m$, and predict $Y_{um}$ as:\n",
      "\n",
      "$$ \\hat{Y_{um}} = \\hat Y^{baseline}_{um}\\, + \\,\\frac{\\sum\\limits_{j \\in S^{k}(m;u)} s_{mj} ( Y_{uj} - \\hat Y^{baseline}_{uj} )}{\\sum\\limits_{j \\in S^{k}(m;u)} s_{mj} } $$\n",
      "\n",
      "where $s^{k}(m;u)$ is the $k$ neighbor items of item $m$ which have been rated by user $u$.\n",
      "\n",
      "Now, this is not a particularly good assumption, especially in the situation where a restaurant is new (new item problem) or a user is new (cold start problem), or in the case when there are very few reviewers of a restaurant, or very few reviews by a user respectively. However, one must start somewhere!\n",
      "\n",
      "Notice that in adding in the similarity term, we subtract the baseline estimate from the observed rating of the user's neighbor items."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Defining the predicted rating"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**2.1** Write a function `knearest_amongst_userrated`, analogous to the `knearest` function we defined above, to find the nearest `k` neighbors to a given restaurant from the **restaurants that the user has already rated**. This function will take as arguments the `restaurant_id`, the `user_id`, the dataframe of reviews, the database, the `k`, and the regularizer `reg`. Just like before, return a *k-length sorted list* of 3-tuples each corresponding to a restaurant.  HINT: use the `knearest` function you defined earlier"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "\"\"\"\n",
      "Function\n",
      "--------\n",
      "knearest_amongst_userrated\n",
      "\n",
      "Parameters\n",
      "----------\n",
      "restaurant_id : string\n",
      "    The id of the restaurant whose nearest neighbors we want\n",
      "user_id : string\n",
      "    The id of the user, in whose reviewed restaurants we want to find the neighbors\n",
      "df: Dataframe\n",
      "    The dataframe of reviews such as smalldf\n",
      "dbase : instance of Database class.\n",
      "    A database of similarities, on which the get method can be used to get the similarity\n",
      "  of two businessed. e.g. dbase.get(rid1,rid2)\n",
      "k : int\n",
      "    the number of nearest neighbors desired, default 7\n",
      "reg: float\n",
      "    the regularization.\n",
      "    \n",
      "  \n",
      "Returns\n",
      "--------\n",
      "A sorted list\n",
      "    of the top k similar restaurants. The list is a list of tuples\n",
      "    (business_id, shrunken similarity, common support).\n",
      "\"\"\"\n",
      "#your code here\n",
      "def knearest_amongst_userrated(restaurant_id, user_id, df, dbase, k=7, reg=3.):\n",
      "    dfuser=df[df.user_id==user_id]\n",
      "    bizsuserhasrated=dfuser.business_id.unique()\n",
      "    return knearest(restaurant_id, bizsuserhasrated, dbase, k=k, reg=reg)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 28
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**2.2** Now write a function that returns the predicted rating for a user and an item using the formula at the beginning of this problem. Include code to deal with the possibility that the sum of scores that goes in the denominator is 0: return an predicted rating of the baseline portion of the formula in that case. This function `rating` takes as arguments the dataframe, the database, the wanted `restaurant_id` and `user_id`, and `k` as well as the regularizer."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "\"\"\"\n",
      "Function\n",
      "--------\n",
      "rating\n",
      "\n",
      "Parameters\n",
      "----------\n",
      "df: Dataframe\n",
      "    The dataframe of reviews such as smalldf\n",
      "dbase : instance of Database class.\n",
      "    A database of similarities, on which the get method can be used to get the similarity\n",
      "  of two businessed. e.g. dbase.get(rid1,rid2)\n",
      "restaurant_id : string\n",
      "    The id of the restaurant whose nearest neighbors we want\n",
      "user_id : string\n",
      "    The id of the user, in whose reviewed restaurants we want to find the neighbors\n",
      "k : int\n",
      "    the number of nearest neighbors desired, default 7\n",
      "reg: float\n",
      "    the regularization.\n",
      "    \n",
      "  \n",
      "Returns\n",
      "--------\n",
      "A float\n",
      "    which is the impued rating that we predict that user_id will make for restaurant_id\n",
      "\"\"\"\n",
      "#your code here\n",
      "def rating(df, dbase, restaurant_id, user_id, k=7, reg=3.):\n",
      "    mu=df.stars.mean()\n",
      "    users_reviews=df[df.user_id==user_id]\n",
      "    nsum=0.\n",
      "    scoresum=0.\n",
      "    nears=knearest_amongst_userrated(restaurant_id, user_id, df, dbase, k=k, reg=reg)\n",
      "    restaurant_mean=df[df.business_id==restaurant_id].business_avg.values[0]\n",
      "    user_mean=users_reviews.user_avg.values[0]\n",
      "    scores=[]\n",
      "    for r,s,nc in nears:\n",
      "        scoresum=scoresum+s\n",
      "        scores.append(s)\n",
      "        r_reviews_row=users_reviews[users_reviews['business_id']==r]\n",
      "        r_stars=r_reviews_row.stars.values[0]\n",
      "        r_avg=r_reviews_row.business_avg.values[0]\n",
      "        rminusb=(r_stars - (r_avg + user_mean - mu))\n",
      "        nsum=nsum+s*rminusb\n",
      "    baseline=(user_mean +restaurant_mean - mu)\n",
      "    #we might have nears, but there might be no commons, giving us a pearson of 0\n",
      "    if scoresum > 0.:\n",
      "        val =  nsum/scoresum + baseline\n",
      "    else:\n",
      "        val=baseline\n",
      "    return val\n",
      "        "
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 29
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "For the top-recommendations in the variable `toprecos` from the previous section, we compute the predicted rating and compare it with the average rating over all users available inside the tuples that make up `toprecos`. We use a `k` of 7 and regularization 3. For comparision we also print this users' average rating. Do you notice anything interesting about how the order has changed from when we did this with the global similarities? (for you to think, not to answer)"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print \"User Average\", smalldf[smalldf.user_id==testuserid].stars.mean(),\"for\",usernamefromid(smalldf,testuserid)\n",
      "print \"Predicted ratings for top choices calculated earlier:\"\n",
      "for biz_id,biz_avg in toprecos:\n",
      "    print biznamefromid(smalldf, biz_id),\"|\",rating(smalldf, db, biz_id, testuserid, k=7, reg=3.),\"|\",\"Average\",biz_avg "
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "User Average 3.5652173913 for Vern\n",
        "Predicted ratings for top choices calculated earlier:\n",
        "Rokerij | 4.71714023074 | Average 4.37931034483\n",
        "Wildfish Seafood Grille | 4.27594504172 | Average 4.29411764706\n",
        "Cornish Pasty Company | 4.62810510121 | Average 4.20689655172\n",
        "Pappadeaux Seafood Kitchen | 4.08845573953 | Average 4.18518518519\n",
        "Four Peaks Brewing Co | 4.26174734161 | Average 4.16666666667\n"
       ]
      }
     ],
     "prompt_number": 30
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Testing the ratings\n",
      "\n",
      "Let us compare the predicted ratings with a user's ratings. Note that we are doing this on the same set that we constructed the predictions with, so this is not a validation of the procedure, but simply a check of the procedure's fit. We first write a helper function to return the user score for a restaurant, and the restaurant's average score over all users."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def get_other_ratings(restaurant_id, user_id, df):\n",
      "    \"get a user's rating for a restaurant and the restaurant's average rating\"\n",
      "    choice=df[(df.business_id==restaurant_id) & (df.user_id==user_id)]\n",
      "    users_score=choice.stars.values[0]\n",
      "    average_score=choice.business_avg.values[0]\n",
      "    return users_score, average_score"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 31
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "For the user `testuserid`, we loop over the variable `bizs` (which is a set of restaurants the user has rated) and print the predicted rating, and the actual rating and restaurant average rating obtained using the function above. We again use `k=7` and a regularization of 3."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print \"for user\",usernamefromid(smalldf,testuserid), 'avg', smalldf[smalldf.user_id==testuserid].stars.mean() \n",
      "for biz_id in bizs:\n",
      "    print \"----------------------------------\"\n",
      "    print biznamefromid(smalldf, biz_id)\n",
      "    print \"Predicted Rating:\",rating(smalldf, db, biz_id, testuserid, k=7, reg=3.) \n",
      "    u,a=get_other_ratings(biz_id, testuserid, smalldf)\n",
      "    print \"Actual User Rating:\",u,\"Avg Rating\",a"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "for user Vern avg 3.5652173913\n",
        "----------------------------------\n",
        "Local Breeze\n",
        "Predicted Rating: 4.2280987611\n",
        "Actual User Rating: 5 Avg Rating 4.0\n",
        "----------------------------------\n",
        "Carly's Bistro\n",
        "Predicted Rating: 3.99008654065\n",
        "Actual User Rating: 5 Avg Rating 3.5\n",
        "----------------------------------\n",
        "Tee Pee Mexican Food\n",
        "Predicted Rating: 3.52640184162\n",
        "Actual User Rating: 5 Avg Rating 3.04347826087\n",
        "----------------------------------\n",
        "District American Kitchen and Wine Bar\n",
        "Predicted Rating: 3.80281696528\n",
        "Actual User Rating: 4 Avg Rating 3.55263157895\n",
        "----------------------------------\n",
        "Los Reyes de la Torta\n",
        "Predicted Rating: 3.41514298977\n",
        "Actual User Rating: 4 Avg Rating 4.13157894737\n"
       ]
      }
     ],
     "prompt_number": 32
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**2.3** Explain in words why the predicted ratings are lower than the actual ratings. How do the user average rating and restaurant average rating affect this? How does sparsity affect the predicted ratings?"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*your answer here*\n",
      "\n",
      "Recall that bizs (defined just above question 1.7) has restaurants sorted by Vern's actual star ratings. This means that in this sample, we are looking at Vern's top rated restaurants.\n",
      "\n",
      "The predicted ratings are lower because these are Vern's top 5 choices, which represent the largest positive deviations away from Vern's mean rating of 3.57. Because we are looking at the upper tail of Vern's rating distribution, but pooling information together from the K nearest neighbors among Vern's rated restaurants to construct the predicted rating, the predicted ratings should fall closer to Vern's user mean than the true ones do. Taking into account the average restaurant rating helps a little bit here because we can adjust the predicted rating to reflect an overall very good restaurant, but it does not counteract the effect of looking at the upper tail of Vern's ratings.\n",
      "\n",
      "Note that if we were to take Vern's bottom 5 restaurants, we would see the opposite effect.\n",
      "\n",
      "In general, the larger K is (assuming that the similarities within this neighborhood are positive), the closer the predicted rating will be to Vern's user average (this is the bias limit in the bias-variance tradeoff). Similarly, the smaller K is, the more likely we are to have user ratings that are close to the observed rating (the variance limit). The sparsity of the data affects how quickly we move from the variance limit to the bias limit as we increase K. If there were a lot of very similar restaurants in the dataset that Vern had ranked very highly, even with K relatively large, it would be possible to see a predicted rating much closer to the extremely positive ratings we see here in Vern's top 5 (see the results in question 4.4). As these data are now, however, even the most similar 7 restaurants to these that Vern rated so highly lie closer to Vern's mean."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Error Analysis\n",
      "\n",
      "This next function takes a set of actual ratings, and a set of predicted ratings, and plots the latter against the former. We can use a graph of this kind to see how well or badly we do in our predictions. Since the nearest neighbor models can have alternating positive and negative similarities (the sum of similarity weights in the denominator can get large), the ratings can get very large. Thus we restrict ourselves to be between -10 and 15 in our ratings and calculate the fraction within these bounds. We also plot the line with unit slope, line sehments joining the means, and a filled in area representing one standard deviation from the mean.\n",
      "\n",
      "The first argument to `compare_results` is a numpy array of the actual star ratings obtained from the dataframe, while the second argument is the numpy array of the predicted ones. (*Feel free to improve this function for your display*)"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def compare_results(stars_actual, stars_predicted, ylow=-10, yhigh=15, title=\"\"):\n",
      "    \"\"\"\n",
      "    plot predicted results against actual results. Takes 2 arguments: a\n",
      "    numpy array of actual ratings and a numpy array of predicted ratings\n",
      "    scatterplots the predictions, a unit slope line, line segments joining the mean,\n",
      "    and a filled in area of the standard deviations.\"\n",
      "    \"\"\"\n",
      "    fig=plt.figure()\n",
      "    df=pd.DataFrame(dict(actual=stars_actual, predicted=stars_predicted))\n",
      "    ax=plt.scatter(df.actual, df.predicted, alpha=0.2, s=30, label=\"predicted\")\n",
      "    plt.ylim([ylow,yhigh])\n",
      "    plt.plot([1,5],[1,5], label=\"slope 1\")\n",
      "    xp=[1,2,3,4,5]\n",
      "    yp=df.groupby('actual').predicted.mean().values\n",
      "    plt.plot(xp,yp,'k', label=\"means\")\n",
      "    sig=df.groupby('actual').predicted.std().values\n",
      "    plt.fill_between(xp, yp - sig, yp + sig, \n",
      "                 color='k', alpha=0.2)\n",
      "    plt.xlabel(\"actual\")\n",
      "    plt.ylabel(\"predicted\")\n",
      "    plt.legend(frameon=False)\n",
      "    remove_border()\n",
      "    plt.grid(False)\n",
      "    plt.title(title)\n",
      "    print \"fraction between -15 and 15 rating\", np.mean(np.abs(df.predicted) < 15)\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 33
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**2.4** For each review in the data set, obtain a prediction from the entire dataframe `smalldf`. Use the function `compare_results` above to plot the predicted ratings against the observed ones. Make 4 such graphs, at k=3 and k=10, and for reg=3. and reg=15.\n",
      "\n",
      "Note that this analysis is not strictly a model check because we are testing on the training set. However, since the user averages would change each time a cross-validation split was done on the set, we would incur the prohibitive expense of redoing the database each time. This would be better done on a cluster, using map-reduce or other techniques. While we explore map-reduce later in this homework, we shall not do any cross-validation.\n",
      "\n",
      "Explain the results you get in the graphs in words."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "def make_results_plot(df,k,reg):\n",
      "    uid=smalldf.user_id.values\n",
      "    bid=smalldf.business_id.values\n",
      "    actual=smalldf.stars.values\n",
      "    predicted=np.zeros(len(actual))\n",
      "    counter=0\n",
      "    for user_id, biz_id in zip(uid,bid):\n",
      "        predicted[counter]=rating(smalldf, db, biz_id, user_id, k=k, reg=reg) \n",
      "        counter=counter+1\n",
      "    compare_results(actual, predicted)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 34
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "print \"k=3, reg=3.\"\n",
      "make_results_plot(smalldf,3,3.)\n",
      "plt.title(\"k=3, reg=3.\")\n",
      "\n",
      "print \"k=3, reg=15.\"\n",
      "make_results_plot(smalldf,3,15.,)\n",
      "plt.title(\"k=3, reg=15.\")\n",
      "\n",
      "print \"k=10, reg=3.\"\n",
      "make_results_plot(smalldf,10,3.)\n",
      "plt.title(\"k=10, reg=3.\")\n",
      "\n",
      "print \"k=10, reg=15.\"\n",
      "make_results_plot(smalldf,10,15.,)\n",
      "plt.title(\"k=10, reg=15.\")"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "k=3, reg=3.\n",
        "fraction between -15 and 15 rating"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.0\n",
        "k=3, reg=15.\n",
        "fraction between -15 and 15 rating"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.999837793998\n",
        "k=10, reg=3.\n",
        "fraction between -15 and 15 rating"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.996431467964\n",
        "k=10, reg=15.\n",
        "fraction between -15 and 15 rating"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 0.997080291971\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 35,
       "text": [
        "<matplotlib.text.Text at 0x10cec8250>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmwAAAGRCAYAAADYce9/AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10cdee410>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmwAAAGSCAYAAABe5Z3RAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x107be4910>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmwAAAGRCAYAAADYce9/AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x107be4650>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmwAAAGSCAYAAABe5Z3RAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nPpy2AAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10ce93990>"
       ]
      }
     ],
     "prompt_number": 35
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*your answer here*\n",
      "\n",
      "If you are a bit confused by the look of these graphs, you should be!\n",
      "\n",
      "For k=3, the predicted values are quite well-behaved, with the exception of several predictions that have extremely large magnitudes. It appears that the predicted values are pulled into the mean star rating, which sits somewhere around 3.8, so ratings on the low end are overestimated, and similarly ratings on the high end underestimated. The regularization does not appear to have a strong effect when k = 3.\n",
      "\n",
      "For k=10, the predicted values are much less stable, with many more extreme predictions. The means appear to track better with the true means. The regularization has a much more extreme, although indirect, effect on the appearance of the plot. Since regularization has stronger effects on similarity scores between restaurants that have small common support, we can see that increasing k makes the predictions more sensitive to the regularization because in a small dataset, the common support between a restaurant and it's 10-nearest one will be quite small.\n",
      "\n",
      "Note that this example does not seem to follow the standard bias-variance tradeoff, where we would expect small k to give a unbiased estimates that capture the extremes, while we would expect large k to give biased estimates that pull extreme values toward the mean. A large reason for this failure for this example to capture this behavior is that we have defined similarity scores that can be positive or negative, and the bias-variance logic is based on the more standard setting where we average together values that have strictly positive weights. When you have negative weights, it's possible that the sum of $s_{ij}$'s in a neighborhood can get close to zero, making our estimator $\\hat Y_{um}$ explode in the positive or negative direction since this would entail dividing by (nearly) zero. Thus for those restaurants where the denominator goes to 0 (more likely to happen at larger k as you have more chances of it there being more weights to add), the ratings are unstable, even numerically!\n",
      "\n",
      "This problem is less pronounced in large datasets or with small k because the k-nearest restaurants are likely to have positive similarity with the current one. However, with small datasets, we can find that even with k relatively small (in this case, around 10), there are negative similarities in the k-neighborhood that make the estimator unstable. This sort of instability would be much less pronounced in the large dataset.\n",
      "\n",
      "If we were to rescale the similarities to be positive, say between 0 and 1, the behavior of the estimator with respect to $k$ and $reg$ would be quite different. (SEE BELOW!)"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "### IMPORTANT SOLUTION ADDENDUM\n",
      "\n",
      "The wide wings we see above are in the error graph are due to small values of the denominator in the similarity sum. If you were writing a production recommender, using the ratings as is from Q2 would be a very bad idea. \n",
      "\n",
      "Indeed the very idea of nearest neighbor should engender in you the idea of distances; and distances should not be negative.\n",
      "Furthermore, you would not be able to use distances other than those implied by the correlation coefficient in the calculation of nearest neighbors (such as a Manhattan distance or Jacard similarity).\n",
      "\n",
      "We can fix this for the case of perarson coefficient by just a simple rescaling to the 0-1 range!\n",
      "\n",
      "$$ \\rho \\rightarrow \\frac{\\rho+1}{2} $$.\n",
      "\n",
      "This translates into changing the shrunk score:\n",
      "\n",
      "$$ s \\rightarrow \\frac{s}{2} + \\frac{f}{2} $$\n",
      "\n",
      "where \n",
      "\n",
      "$$ f = \\frac{N_{common}}{N_{common}+reg} $$\n",
      "\n",
      "Note that the new quantity is really an inverse distance, and thus we'll sort by its smallness as before."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def knearest_pos(restaurant_id, set_of_restaurants, dbase, k=7, reg=3.):\n",
      "    \"\"\"\n",
      "    Given a restaurant_id, dataframe, and database, get a sorted list of the\n",
      "    k most similar restaurants from the entire database.\n",
      "    \"\"\"\n",
      "    similars=[]\n",
      "    for other_rest_id in set_of_restaurants:\n",
      "        if other_rest_id!=restaurant_id:\n",
      "            sim, nc=dbase.get(restaurant_id, other_rest_id)\n",
      "            ssim=shrunk_sim(sim, nc, reg=reg)\n",
      "            similars.append((other_rest_id, ssim/2.0 + float(nc)/(float(nc)+reg), nc ))\n",
      "    similars=sorted(similars, key=itemgetter(1), reverse=True)\n",
      "    return similars[0:k]\n",
      "\n",
      "def knearest_amongst_userrated_pos(restaurant_id, user_id, df, dbase, k=7, reg=3.):\n",
      "    dfuser=df[df.user_id==user_id]\n",
      "    bizsuserhasrated=dfuser.business_id.unique()\n",
      "    return knearest_pos(restaurant_id, bizsuserhasrated, dbase, k=k, reg=reg)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 36
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def rating_pos(df, dbase, restaurant_id, user_id, k=7, reg=3.):\n",
      "    mu=df.stars.mean()\n",
      "    users_reviews=df[df.user_id==user_id]\n",
      "    nsum=0.\n",
      "    scoresum=0.\n",
      "    nears=knearest_amongst_userrated_pos(restaurant_id, user_id, df, dbase, k=k, reg=reg)\n",
      "    restaurant_mean=df[df.business_id==restaurant_id].business_avg.values[0]\n",
      "    user_mean=users_reviews.user_avg.values[0]\n",
      "    scores=[]\n",
      "    for r,sold,nc in nears:\n",
      "        s=sold/2.0\n",
      "        shrink_factor=float(nc)/(float(nc)+reg)\n",
      "        s=s+shrink_factor/2.0\n",
      "        scoresum=scoresum+s\n",
      "        scores.append(s)\n",
      "        r_reviews_row=users_reviews[users_reviews['business_id']==r]\n",
      "        r_stars=r_reviews_row.stars.values[0]\n",
      "        r_avg=r_reviews_row.business_avg.values[0]\n",
      "        rminusb=(r_stars - (r_avg + user_mean - mu))\n",
      "        nsum=nsum+s*rminusb\n",
      "    baseline=(user_mean +restaurant_mean - mu)\n",
      "    #we might have nears, but there might be no commons, giving us a pearson of 0\n",
      "    if scoresum > 0.:\n",
      "        val =  nsum/scoresum + baseline\n",
      "    else:\n",
      "        val=baseline\n",
      "    return val"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 37
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def make_results_plot_pos(df,k,reg):\n",
      "    uid=smalldf.user_id.values\n",
      "    bid=smalldf.business_id.values\n",
      "    actual=smalldf.stars.values\n",
      "    predicted=np.zeros(len(actual))\n",
      "    counter=0\n",
      "    for user_id, biz_id in zip(uid,bid):\n",
      "        predicted[counter]=rating_pos(smalldf, db, biz_id, user_id, k=k, reg=reg) \n",
      "        counter=counter+1\n",
      "    compare_results(actual, predicted, ylow=1, yhigh=5)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 38
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "NOTICE: when comparing graphs note that the limits have changed to make the bias-variance comparison more visible. The regularizer is set to 1 to reduce the regularization to something small, rather than something optimal, as well. Also to make things more extreme, I choose k=15 rather than k=10."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print \"k=2, reg=1.\"\n",
      "make_results_plot_pos(smalldf,2,1.)\n",
      "plt.title(\"k=2, reg=1.\")\n",
      "\n",
      "print \"k=2, reg=15.\"\n",
      "make_results_plot_pos(smalldf,2,15.,)\n",
      "plt.title(\"k=2, reg=15.\")\n",
      "\n",
      "print \"k=15, reg=1.\"\n",
      "make_results_plot_pos(smalldf,15,1.)\n",
      "plt.title(\"k=15, reg=1.\")\n",
      "\n",
      "print \"k=15, reg=15.\"\n",
      "make_results_plot_pos(smalldf,15,15.,)\n",
      "plt.title(\"k=15, reg=15.\")"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "k=2, reg=1.\n",
        "fraction between -15 and 15 rating"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.0\n",
        "k=2, reg=15.\n",
        "fraction between -15 and 15 rating"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.0\n",
        "k=15, reg=1.\n",
        "fraction between -15 and 15 rating"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.0\n",
        "k=15, reg=15.\n",
        "fraction between -15 and 15 rating"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.0\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 39,
       "text": [
        "<matplotlib.text.Text at 0x10cef8990>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmYAAAGRCAYAAADPU3+2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n+dgAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10cdfdcd0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmYAAAGSCAYAAABJxw0YAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nERFRkGBhRkRERBQkWJgRERERBYn/D0j1ulZpqd7PAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10ce1e250>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmYAAAGSCAYAAABJxw0YAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMwAAAB/xf3DDi13H8HyGAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10cf0bc90>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmYAAAGSCAYAAABJxw0YAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nHgQqpHEAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10ce62590>"
       ]
      }
     ],
     "prompt_number": 39
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Notice firstly, that at low regularization, at low k (variance limit), the predictions follow the green line more than at high k ( bias limit). This is, as mentioned earlier, due to the bias limit pulling the extreme groups of ratings in towards the mean of all the reviews. Note that increasing bias decreases the variation **between** the rating groups, resulting in a flatter black curve.\n",
      "\n",
      "The second thing to notice that as k goes up, the precision as measured by the width of the grey region (which measures the variance **WITHIN** rating groups, not between them) gets smaller. \n",
      "\n",
      "The third thing to notice is that adding in regularization in the variance (low k) limit actually increases the bias, which is not surprising as it tends to push things towards the center. In many machine learning situations this is a desirable thing, as it reduces overfitting. In the high bias limit, as expected, it does not have much of an impact. Note that regularization had more of an impact in the previous set of graphs, as it pushed similarity towards 0.\n",
      "\n",
      "One might ask: why not plot these graphs in the homework in the first place: using distances is the correct way to do it, and dosent lead to the subtraction problem. Its a good question, and you should use positive similarities in a production recommender. For the puposes of this homework, i think the k=10 graphs in the previous set illustrate the problem with sparsity in a much more dramatic way, leading to the \"bellows\" in the graph. However, in real life, there's no real justification for using similarities that can be negative as weights. If we think of KNN as a way of constructing local averages, you need those averages to be constructed using positive weights. "
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**2.5** Outline a process, in words, for choosing the nearest neighbor parameter `k`. For this question fix the regularization parameter `reg` at `3`."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*your answer here*\n",
      "\n",
      "We could use $F$-fold cross-validation (usually called $k$-fold cross-validation, but we've already used $k$ here to mean something else!). Specifically, we could randomly partition the data into $F$ equally sized folds (pieces), making sure that each fold includes at least one rating for each restaurant and one rating by each user (it would probably best to make sure that if $K$ were the maximum $k$ you would be considering, that each user and each restaurant appear at least $K$ times in each fold). For each value of $k$, and for each fold, we could repeat the procedure above for predicting user ratings by computing similarities using $F-1$ of the folds to compute similarities and computing the prediction error in the held out fold. We could then choose the $k$ with the smallest average prediction error across the folds, and recompute the recommender on the whole dataset using the chosen value of $k$.\n",
      "\n",
      "If we wanted to both choose a good value for $k$ and check how well we could expect the result to generalize, we could divide the dataset into $F+1$ folds, and keep the last fold out as a verification set. We could perform the cross-validation above on $F$ folds to select $k$, then upon selecting $k$ use all $F$ folds to create a recommender, and see how well this recommender predicted ratings in the final validation set."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "##Q3 Bayesian Chocolates: Model based recommendations"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "In this part of the homework, you will use your newly minted Bayesian and Gibbs sampler skills to write a recommender that uses Bayesian techniques to impute ratings."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Model-Based Recommendations\n",
      "\n",
      "\n",
      "####A Note on Frequentist and Bayesian Procedures\n",
      "\n",
      "In the previous section we implemented a **procedure** (a set of instructions for processing data) for giving recommendations and predicting user ratings for restaurants. This procedure involved a number of arbitrary choices -- for example, the particular measure of similarity between restaurants, or the weighting scheme for constructing a predicted rating. It also gave no sense of uncertainty -- in the case of giving recommendations, there was no statement about how we would expect the ranking from the procedure to compare to the user's true opinions of restaurants, and in the case of predicting ratings, there was no confidence interval for the prediction.\n",
      "\n",
      "It is possible in repeated applications of the above procedure to see how it performs in the long run. Based on this long-run performance we could potentially justify certain functional choices and compute measurements of uncertainty. This framework of proposing a procedure first, then evaluating its performance in real or hypothetical replications of the experiment is an example of a *frequentist* approach to a problem. One aspect of the frequentist approach is that the proposed procedure does not necessarily have to be derived from a model (although it often is). While this means that a proposed procedure may be more flexible or robust than a model-based procedure, it also means that there is no natural way to justify certain functional choices or construct uncertainty estimates.\n",
      "\n",
      "In contrast, the *Bayesian* approach to a problem always begins with a **probablistic model** for how the data were generated. Assuming this model is true, the posterior distribution over unknown quantities (either parameters to be estimated or unobserved data to be predicted) gives a single coherent expression of what the observed data tell us about the unknowns. By summarizing the posterior distribution, we can derive the exact functional form of a procedure for constructing estimates or predictions. We call a procedure derived from this Bayesian approach a **Bayes rule** (not to be confused with Bayes' Theorem). Using the posterior distribution, we can also give a sense of how uncertain we are about the estimate or prediction we have constructed.\n",
      "\n",
      "####Outline for this Problem\n",
      "\n",
      "In this section, we construct a **model** of how ratings are generated, and use this model to build a recommendation and ratings prediction system. We will take a Bayesian approach here, and construct our estimates and predictions from summaries of the *posterior distribution* of the model's parameters, which we will compute using a *Gibbs sampler*. We will also give measures of uncertainty based on the posterior distribution. We will evaluate predictions from this approach in the same way we evalutated predictions from the KNN procedure above."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###The Latent Factor Model###\n",
      "\n",
      "####Model Overview####\n",
      "\n",
      "The central dogma in constructing a recommendation system using collaborative filtering is that *similar users will rate similar restaurants similarly*. In the previous section, we explicitly encoded this idea by using a similarity function to identify similar restaurants. We also assumed that either all users were the same (the global approach) or that only the current user was similar enough to make a recommendation (the user-specific approach). In this section, we will use a model that allows us to identify both similar users and similar restaurants as a function of **latent factors**.\n",
      "\n",
      "We can think of latent factors as properties of restaurants (e.g., spiciness of food or price) that users have a positive or negative preference for. We do not observe these factors or the users' preferences directly, but we assume that they affect how users tend to rate restaurants. For example, if a restaurant serves a lot of spicy food and a user dislikes spicy food, then the restaurant would have a high \"spiciness\" factor, and the user would have a strongly negative preference, resulting in a prediction of a low rating. Note that if users have similar preferences, then according to the model, they will behave similarly, and likewise, if restaurants have similar latent factors, they will be rated similarly by similar users. Latent factors thus give us an intuitive way to specify a generative model the obeys the central dogma.\n",
      "\n",
      "One issue that comes up with latent factor models is determining how many latent factors to include. There may be a number of different unmeasured properties that affect ratings in different ways -- for example, in addition to the spiciness factor above, there may also be a price factor that affects how users rate a restaurant. We deal with the problem of choosing the number of latent factors to include in the same way we deal with choosing $K$ in a $K$-nearest neighbors problem.\n",
      "\n",
      "####Rating Model Specification####\n",
      "\n",
      "To make this model concrete, we can write down our probability model as a generative process. First, we define the following quantities:\n",
      "\n",
      "Counts:\n",
      "\n",
      "* $L$: The number of latent factors.\n",
      "\n",
      "* $U$: The number of users.\n",
      "\n",
      "* $M$: The number of items (restaurants).\n",
      "\n",
      "* $N$: The number of observed ratings.\n",
      "\n",
      "Data:\n",
      "\n",
      "* $Y_{um}$: The star rating given to restaurant $m$ by user $u$.\n",
      "* $Y$: The full collection of observed star ratings.\n",
      "\n",
      "Item-specific quantities:\n",
      "\n",
      "* $\\gamma_m$: An item-specific parameter vector of length $L+1$. The first element of $\\gamma_m$, denoted $\\gamma_m[0]$ is the item-specific bias. The remaining $L$ elements of $\\gamma_m$, denoted $\\gamma_m[1:]$, are the latent factors associated with item $m$.\n",
      "\n",
      "* $\\Gamma$: An $M$ by $L+1$ matrix where the $m$th row is $\\gamma_m$.\n",
      "\n",
      "User-specific quantities:\n",
      "\n",
      "* $\\theta_u$: A user-specific parameter vector of length $L+1$. The first element of $\\theta_u$, denoted $\\theta_u[0]$ is the user-specific bias. The remaining $L$ elements of $\\theta_u$, denoted $\\theta_u[1:]$, are user $u$'s preferences for the latent factors.\n",
      "\n",
      "* $\\Theta$: A $U$ by $L+1$ matrix where the $u$th row is $\\theta_u$.\n",
      "\n",
      "Global quantities:\n",
      "\n",
      "* $\\mu$: The overall ratings mean.\n",
      "\n",
      "* $\\sigma$: The residual variance of ratings after the mean, bias terms, and latent factors have been taken into account.\n",
      "\n",
      "Using these quantities, we can specify our model for each rating $Y_{um}$ similarly to a linear regression:\n",
      "\n",
      "$$Y_{um} = \\mu + \\theta_{u}[0] + \\gamma_{m}[0] + \\theta_{u}[1:]^{\\top}\\gamma_{m}[1:] + \\epsilon_{um}$$\n",
      "\n",
      "where\n",
      "\n",
      "$$\\epsilon_{um} \\sim N(0, \\sigma).$$\n",
      "\n",
      "Note that while this looks like a linear regression, it is of a slightly different form because the latent factor term involves the product of two unknowns. This is like a linear regression where we forgot to measure some covariates.\n",
      "\n",
      "We also assume the following priors on the user-specific and item-specific parameters:\n",
      "\n",
      "$$\n",
      "\\begin{align*}\n",
      "\\gamma_m &\\sim MVN(\\mathbf 0, \\Lambda_\\gamma^{-1})\\\\\n",
      "\\theta_u &\\sim MVN(\\mathbf 0, \\Lambda_\\theta^{-1}),\n",
      "\\end{align*}\n",
      "$$\n",
      "\n",
      "where $MVN$ means multivariate normal, $\\mathbf 0$ is vector of length $L+1$ filled with zeros, and $\\Lambda_\\theta^{-1}$ and $\\Lambda_\\gamma^{-1}$ are $L+1 \\times L+1$ covariance matrices. $\\mu$ and $\\sigma$ also have priors, but they are not relevant to your task so we won't write them here.\n",
      "\n",
      "#### Goal for this Model####\n",
      "Using this model, we want to make inference about all of the quantities that, if we knew them, would allow us to sample $Y_{um}$ for any user and any item. These quantities are $\\mu$, $\\sigma$, and the elements of $\\Theta$ and $\\Gamma$."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**3.1**: Given the goal specified above, how many quantities (counting a vector of $L$ items as $L$ quantities) are we trying to make inference about? Express your answer in terms of the variables in the \"Counts\" section above."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*your answer here*\n",
      "\n",
      "There are $$U \\times (L+1) + M \\times (L+1) + 2$$ quantities to estimate."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Gibbs Sampling from the Posterior###\n",
      "\n",
      "Our goal is to compute the **posterior distribution** over the unknowns $\\mu$, $\\sigma$, $\\Gamma$, and $\\Theta$ given $Y$, which reflects how much we know about these quantities given the data we have observed. We write this distribution as $P(\\mu, \\sigma, \\Gamma, \\Theta \\mid Y)$.\n",
      "\n",
      "The most general way to learn about the posterior distribution is to sample from it. This can be challenging, particularly in problems that are very high dimensional (see your answer to the question above). One strategy for for sampling from high-dimensional distributions is **Gibbs sampling**, which we discussed in class and lab.\n",
      "\n",
      "Gibbs sampling breaks down the posterior probability distribution into blocks of unknowns, and samples iteratively from each block assuming that the values of the other blocks (and the data) are known and fixed. In this case, we will break down the posterior distribution into blocks of $\\mu$, $\\sigma$, each vector $\\gamma_m$, and each vector $\\theta_u$. We have already implemented the draws for $\\mu$ and $\\sigma$. You will need to implement the draws for each $\\gamma_m$ and each $\\theta_u$. Luckily, the structures of these draws are similar, so you will only need to implement two functions.\n",
      "\n",
      "First, we'll derive the form of the draws below. Note that you don't need to be able to follow these derivations fully -- you'll just need to be able to use the result at the end.\n",
      "\n",
      "####Distribution of $\\gamma_{m'}$ given $Y, \\mu, \\sigma, \\Gamma_{-m'}, \\Theta$####\n",
      "\n",
      "Intuitively, this is the distribution of the item-specific parameters for item $m'$, imagining that all of the other unknowns are fixed.\n",
      "\n",
      "More precisely, we want to draw from the distribution of $\\gamma_{m'}$ conditional on the data $Y$ and all other unknowns -- that is, $\\mu$, $\\sigma$, all of $\\Theta$, and all of $\\Gamma$ except for $\\gamma_{m'}$, which we denote $\\Gamma_{-m}$.\n",
      "\n",
      "Note that in the model specification above, the only places that $\\gamma_{m'}$ appears are in the regression equations for each $Y_{um}$ that involves item $m'$. If we write out just these equations, we get a system of the following form,\n",
      "\n",
      "$$Y_{um'} = \\mu + \\theta_{u}[0] + \\gamma_{m'}[0] + \\theta_{u}[1:]^{\\top}\\gamma_{m'}[1:] + \\epsilon_{um'},$$\n",
      "\n",
      "with one equation for each $u$ that rated item $m'$. Now, because \n",
      "\n",
      "If we move all of the fully known terms to the left-hand side, we obtain the system:\n",
      "\n",
      "$$Y_{um'} - \\mu - \\theta_{u}[0] = \\gamma_{m'}[0] + \\theta_{u}[1:]^{\\top}\\gamma_{m'}[1:] + \\epsilon_{um'}.$$\n",
      "\n",
      "Notice that, because we assume that $\\theta_{u}$ is known, this equation now fits cleanly into the form of a linear regression, where $\\gamma_{m'}$ is the vector of unknown coefficients. This means that the posterior distribution for $\\gamma_{m'}$ conditional on everything else is the same as the posterior for the coefficients of a Bayesian linear regression of $(Y_{um'} - \\mu - \\theta_{u}[0])$ on $\\theta_{u}[1:]$ and an intercept.\n",
      "\n",
      "Let's denote the set of users who rated item $m'$ as $(u_1, \\cdots, u_g)$. Then, we can define the following vector and matrix:\n",
      "\n",
      "\\begin{align*}\n",
      "Y_{m'} = \\left(\\begin{array}{c} Y_{u_1m'}-\\mu-\\theta_{u_1}[0]\\\\ \\vdots \\\\ Y_{u_gm'}-\\mu-\\theta_{u_g}[0]\\end{array}\\right), \\qquad\n",
      "X_{m'} &= \\left(\\begin{array}{cc} 1 & \\theta_{u_1}[1:]^\\top \\\\ \\vdots & \\vdots \\\\ 1 & \\theta_{u_g}[1:]^\\top\\end{array}\\right),\n",
      "\\end{align*}\n",
      "\n",
      "where $Y_{m'}$ is a vector of length $g$ and $X_{m'}$ is a $g \\times L+1$ matrix.\n",
      "\n",
      "The draw from $\\gamma_{m'}$ given everything else then has the form:\n",
      "$$ \\gamma_{m'} \\mid Y, \\mu, \\sigma, \\Gamma_{-m'}, \\Theta \\sim MVN\\left(Q_{m'}^{-1} \\frac{1}{\\sigma^2}X_{m'}^\\top Y_{m'}, Q_{m'}^{-1}\\right)$$\n",
      "where\n",
      "$$ Q_{m'} = \\left(\\frac{1}{\\sigma^2}X_{m'}^\\top X_{m'} + \\Lambda_\\gamma\\right).$$\n",
      "\n",
      "#### Distribution of $\\theta_{u'}$ given $Y, \\mu, \\sigma, \\Gamma, \\Theta_{-u'}$####\n",
      "\n",
      "Intuitively, this is the distribution of the user-specific parameters for user $u'$, imagining that all of the other unknowns are fixed.\n",
      "\n",
      "We can use a very similar argument to the one above. We can denote the set of items rated by user $u'$ as $(m_1, \\cdots, m_g)$ and define the vector and matrix:\n",
      "\\begin{align*}\n",
      "Y_{u'} = \\left(\\begin{array}{c} Y_{u'm_1}-\\mu-\\gamma_{m_1}[0] \\\\ \\vdots \\\\ Y_{u'm_g}-\\mu-\\gamma_{m_g}[0]\\end{array}\\right), \\qquad\n",
      "X_{u'} &= \\left(\\begin{array}{cc} 1 & \\gamma_{m_1}[1:]^\\top \\\\ \\vdots & \\vdots \\\\ 1 & \\gamma_{m_g}[1:]^\\top\\end{array}\\right),\n",
      "\\end{align*}\n",
      "\n",
      "where $Y_{u'}$ is a vector of length $g$ and $X_{u'}$ is a $g \\times L+1$ matrix.\n",
      "\n",
      "the draw from $\\theta_{u'}$ given everything else has the form:\n",
      "$$ \\theta_{u'} \\mid Y, \\mu, \\sigma, \\Gamma, \\Theta_{-u'} \\sim MVN\\left(Q_{u'}^{-1} \\frac{1}{\\sigma^2}X_{u'}^\\top Y_{u'}, Q_{u'}^{-1}\\right)$$\n",
      "where\n",
      "$$ Q_{u'}= \\left(\\frac{1}{\\sigma^2}X_{u'}^\\top X_{u'} + \\Lambda_\\theta\\right).$$"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**3.2** We will only ask you to implement a tiny portion of the Gibbs sampler. Complete the following functions that implement the conditional posterior draws for $\\gamma_m$ and $\\theta_u$ derived above.\n",
      "\n",
      "**Hint**: `np.random.multivariate_normal` is a good function to know."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "\"\"\"\n",
      "Function\n",
      "--------\n",
      "gamma_m_draw\n",
      "\n",
      "Draw a single sample from the conditional posterior distribution\n",
      "of gamma_m.\n",
      "\n",
      "Inputs\n",
      "-------\n",
      "X_m: A g-by-L+1 matrix, defined above. \n",
      "Y_m: A 1D vector of length g, defined above.\n",
      "sig2: Residual _variance_, as defined above.\n",
      "Lambda_gamma: Prior precision matrix.\n",
      "\n",
      "Outputs\n",
      "--------\n",
      "Single draw from conditional posterior, defined above.\n",
      "\"\"\"\n",
      "#Item-specific parameters given all else\n",
      "#your code here\n",
      "def gamma_m_draw(X_m, Y_m, sig2, Lambda_gamma):\n",
      "\n",
      "    #Compute matrices that define conditional posterior.\n",
      "    Q_m_inv = np.linalg.inv(np.dot(X_m.T, X_m)/sig2+Lambda_gamma)\n",
      "    XtY = np.dot(X_m.T, Y_m)\n",
      "\n",
      "    #Draw item-specific parameters.\n",
      "    return np.random.multivariate_normal(np.dot(Q_m_inv, XtY)/sig2, Q_m_inv)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 40
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "\"\"\"\n",
      "Function\n",
      "--------\n",
      "theta_u_draw\n",
      "\n",
      "Draw a single sample from the conditional posterior distribution\n",
      "of gamma_m.\n",
      "\n",
      "Inputs\n",
      "-------\n",
      "X_u: A g-by-L+1 matrix, defined above. \n",
      "Y_u: A 1D vector of length g, defined above.\n",
      "sig2: Residual _variance_, as defined above.\n",
      "Lambda_theta: Prior precision matrix.\n",
      "\n",
      "Outputs\n",
      "--------\n",
      "Single draw from conditional posterior, defined above.\n",
      "\"\"\"\n",
      "#User-specific parameters given all else\n",
      "#your code here\n",
      "def theta_u_draw(X_u, Y_u, sig2, Lambda_theta):\n",
      "    #Compute matrices that define conditional posterior.\n",
      "    Q_u_inv = np.linalg.inv(np.dot(X_u.T, X_u)/sig2+Lambda_theta)\n",
      "    XtY = np.dot(X_u.T, Y_u)\n",
      "    \n",
      "    #Draw the user-specific parameters\n",
      "    return np.random.multivariate_normal(np.dot(Q_u_inv, XtY)/sig2, Q_u_inv)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 41
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Here is the Gibbs sampler skeleton that your functions fit into. Look over the structure to see how for each draw from the posterior, the sampler iterates through $\\mu$, $\\sigma$, $\\gamma_m$ for each item, and $\\theta_u$ for each user."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "\"\"\"\n",
      "Function\n",
      "--------\n",
      "factor_gibbs\n",
      "\n",
      "Runs a gibbs sampler to infer mean, variance, user-specific, and item-specific\n",
      "parameters.\n",
      "\n",
      "Inputs\n",
      "-------\n",
      "data: A dataframe containing ratings data.\n",
      "L: Dimension of latent factors.\n",
      "maxit: Number of samples to draw from posterior.\n",
      "Lambda_theta_diag: Hyperparameter controlling regularization of Theta.\n",
      "Lambda_gamma_diag: Hyperparameter controlling regularization of Gamma.\n",
      "progress: if true, print iteration number every 100 iterations.\n",
      "\n",
      "Outputs\n",
      "--------\n",
      "Dictionary with elements\n",
      "mu: Draws of mu. 1D array of length maxiter.\n",
      "sig2: Draws of sig2, residual _variance_. 1D array of length maxiter.\n",
      "theta: Draws of Theta. U-by-L-by-maxiter array.\n",
      "gamma: Draws of Gamma. M-by-L-by-maxiter array.\n",
      "EY: Draws of fitted values of Y. N-by-maxiter array.\n",
      "\"\"\"\n",
      "def factor_gibbs(data, L, maxit, Lambda_theta_diag, Lambda_gamma_diag, progress=True):\n",
      "    data = data.copy()\n",
      "    N = data.shape[0]\n",
      "\n",
      "    #Create indices that allow us to map users and restaurants to rows\n",
      "    #in parameter vectors.\n",
      "    uusers, uidx = np.unique(data.user_id, return_inverse=True)\n",
      "    uitems, midx = np.unique(data.business_id, return_inverse=True)\n",
      "\n",
      "    nusers = uusers.size\n",
      "    nitems = uitems.size\n",
      "\n",
      "    #Add numerical indices to dataframe.\n",
      "    data[\"uidx\"] = uidx\n",
      "    data[\"midx\"] = midx\n",
      "\n",
      "    #Group observations by user and by business.\n",
      "    ugroups = data.groupby(\"uidx\")\n",
      "    mgroups = data.groupby(\"midx\")\n",
      "\n",
      "    all_avg = data.stars.mean()\n",
      "    u_avg = ugroups.stars.mean()\n",
      "    m_avg = mgroups.stars.mean()\n",
      "\n",
      "    #Initialize parameters and set up data structures for\n",
      "    #holding draws.\n",
      "    #Overall mean\n",
      "    mu = all_avg\n",
      "    mu_draws = np.zeros(maxit)\n",
      "    #Residual variance\n",
      "    sig2 = 0.5\n",
      "    sig2_draws = np.zeros(maxit)\n",
      "\n",
      "    #Matrix of user-specific bias and L latent factors.\n",
      "    theta = np.zeros([nusers, L+1])\n",
      "    theta[:,0] = u_avg-all_avg\n",
      "    theta_draws = np.zeros([nusers, L+1, maxit])\n",
      "\n",
      "    #Matrix of item-specific bias and L latent factors.\n",
      "    gamma = np.zeros([nitems, L+1])\n",
      "    gamma[:,0] = m_avg-all_avg\n",
      "    gamma_draws = np.zeros([nitems, L+1, maxit])\n",
      "\n",
      "    #Matrix for holding the expected number of stars\n",
      "    #for each observation at each draw from the posterior.\n",
      "    EY_draws = np.zeros([data.shape[0], maxit])\n",
      "\n",
      "    #Inverse covariance matrices from the prior on each theta_u\n",
      "    #and gamma_b. These are diagonal, like Ridge regression.\n",
      "    Lambda_theta = np.eye(L+1)*Lambda_theta_diag\n",
      "    Lambda_gamma = np.eye(L+1)*Lambda_gamma_diag\n",
      "\n",
      "    #Main sampler code\n",
      "    for i in range(maxit):\n",
      "        if i%100==0 and progress:\n",
      "            print i\n",
      "\n",
      "        #The entire regression equation except for the overall mean.\n",
      "        nomu = np.sum(theta[data.uidx,1:]*gamma[data.midx,1:], axis=1) +\\\n",
      "                  theta[data.uidx,0] + gamma[data.midx,0]\n",
      "\n",
      "        #Compute the expectation of each observation given the current\n",
      "        #parameter values.\n",
      "        EY_draws[:,i]=mu+nomu\n",
      "\n",
      "        #Draw overall mean from a normal distribution\n",
      "        mu = np.random.normal(np.mean(data.stars-nomu), np.sqrt(sig2/N))\n",
      "        #Draw overall residual variance from a scaled inverse-Chi squared distribution.\n",
      "        sig2 = np.sum(np.power(data.stars-nomu-mu,2))/np.random.chisquare(N-2)\n",
      "        \n",
      "        #For each item\n",
      "        for mi,itemdf in mgroups:\n",
      "            #Gather relevant observations, and subtract out overall mean and\n",
      "            #user-specific biases, which we are holding fixed.\n",
      "            Y_m = itemdf.stars-mu-theta[itemdf.uidx,0]\n",
      "            #Build the regression design matrix implied by holding user factors\n",
      "            #fixed.\n",
      "            X_m = np.hstack((np.ones([itemdf.shape[0],1]),\n",
      "                             theta[itemdf.uidx,1:]))\n",
      "            gamma[mi,:] = gamma_m_draw(X_m, Y_m, sig2, Lambda_gamma)\n",
      "            \n",
      "        #For each user\n",
      "        for ui,userdf in ugroups:\n",
      "            #Gather relevant observations, and subtract out overall mean and\n",
      "            #business-specific biases, which we are holding fixed.\n",
      "            Y_u = userdf.stars-mu-gamma[userdf.midx,0]\n",
      "            #Build the regression design matrix implied by holding business factors\n",
      "            #fixed.\n",
      "            X_u = np.hstack((np.ones([userdf.shape[0],1]),\n",
      "                             gamma[userdf.midx,1:]))\n",
      "            \n",
      "            theta[ui,:] = theta_u_draw(X_u, Y_u, sig2, Lambda_theta)\n",
      "\n",
      "        #Record draws\n",
      "        mu_draws[i] = mu\n",
      "        sig2_draws[i] = sig2\n",
      "        theta_draws[:,:,i] = theta\n",
      "        gamma_draws[:,:,i] = gamma\n",
      "\n",
      "    return {\"mu\": mu_draws, \"sig2\": sig2_draws,\n",
      "            \"theta\": theta_draws, \"gamma\": gamma_draws,\n",
      "            \"EY\": EY_draws}"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 42
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Posterior Summaries###\n",
      "\n",
      "Once you have posterior draws from the sampler, the most natural thing to do is to compute the **posterior mean** of each quantity you are intersted in. To do this, we simply need to take the average value of each quantity across the samples drawn from the sampler. Before taking the average, however, we will want to ignore the first 20-30% of samples because these correspond the **burnin period**, the time during which the sampler is still looking for the main meat of the distribution."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "####Ok it's time to recommend!"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**3.3** Now that you have the Gibbs sampler, draw 1000 samples from the posterior distribution using a two-dimensional latent factor and prior precisions `Lambda_theta_diag` and `Lambda_gamma_diag` both equal to 0.1.\n",
      "\n",
      "Compute the posterior mean of the fitted values for each $Y_{um}$, eliminating the first 200 samples. Call these the `prediction`.  These constitute our recommendations. True to the bayesian paradigm, we dont just have mean predictions, but entire distributions. But currently we are only interested in the means."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "gibbs_out = factor_gibbs(smalldf, 2, 1000, 0.1, 0.1)\n",
      "burnin = 200\n",
      "predicted=np.mean(gibbs_out['EY'][:,burnin:], axis=1)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "0\n",
        "100"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "200"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "300"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "400"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "500"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "600"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "700"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "800"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "900"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n"
       ]
      }
     ],
     "prompt_number": 43
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Plot the predictions against the observed data.You can use the `compare_results` function defined in the previous section. How do the fitted values compare to those from the KNN procedure?"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "compare_results(smalldf.stars.values, predicted, ylow=1, yhigh=5, title=\"From Gibbs Sampler\")"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "fraction between -15 and 15 rating 1.0\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmYAAAGRCAYAAADPU3+2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEXmJ/wfSetTr5X5yfgAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10ce25f10>"
       ]
      }
     ],
     "prompt_number": 44
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*your answer here*\n",
      "\n",
      "The results from the latent factor model appear to be better-behaved than those from the KNN procedure (with negative similarities) since there are no extreme values. In terms of the non-extreme values from the KNN procedure, the latent factor model appears to make similar predictions to the KNN procedure when k = 3., Specifically, the average prediction line for each class is too compressed toward the total data mean of around 3.8, meaning that we are in the bias limit where we are pooling too much information between ratings. Thus, we are overpredicting low ratings and underpredicting high ratings.\n",
      "\n",
      "(If we compare to the KNN procedure with strictly positive weights, as in the homework addendum, we see again that the recommenders make comparable predictions, although the latent factor model again appears to fit slightly better than at both the bias or variance limits of the KNN procedure).\n",
      "\n",
      "There is also a bias-variance tradeoff here. In this case, it appears that we have proposed a model that is too simple (thus close to the bias limit) because of how the ratings are pulled in toward the data mean. In this case, proposing more latent factors increases the flexibility of the model, and thus moves us toward the variance limit. Thus, the plot suggests that we may want to reduce the bias at the cost of increasing variance, for example by considering a latent factor model with more factors (say, $L = 15$) to obtain a better fit (see ADDENDUM below)."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "#### IMPORTANT SOLUTION ADDENDUM"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "gibbs_out = factor_gibbs(smalldf, 15, 1000, 0.1, 0.1)\n",
      "burnin = 200\n",
      "predicted=np.mean(gibbs_out['EY'][:,burnin:], axis=1)\n",
      "compare_results(smalldf.stars.values, predicted, ylow=1, yhigh=5, title=\"From Gibbs Sampler\")"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "0\n",
        "100"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "200"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "300"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "400"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "500"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "600"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "700"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "800"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "900"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "fraction between -15 and 15 rating"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " 1.0\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAmYAAAGRCAYAAADPU3+2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10ce36610>"
       ]
      }
     ],
     "prompt_number": 45
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The fit here looks much better, both in terms of tracking the green line and in terms of the within-group precision (as measured by the grey band). A model at the variance limit (that is, one that is extremely overfit) with low bias would have a mean line that tracks exactly with the green line and an almost non-existent gray area. To tell whether this plot represents simply a good prediction model or a one that is woefully overfit, we would need to look at out-of-sample data."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "##Q4 Scaling Up"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "All our recommenders suffer from problems having to do with the fact that we subsetted an already sparse user-item matrix. The more items we have, the more items we may find in the vicinity of a given item, and thus we are likely to give a more robust average rating to the given item.\n",
      "\n",
      "In this problem we shall use Amazon Elastic Map-Reduce to tackle the entire user-restaurant matrix. We shall do this in two parts: we'll use MRJob locally on your machine to on the smaller data set to calclate the pearson database, and then we'll tackle the entire data set on Amazon.\n",
      "\n",
      "The larger set has 35000 users and 4500 items. Computing the 4500X4500 similarity matrix on one machine will be prohibitively expensive. Thus we'll adopt a strategy where we'll split the calculation over multiple machines using the map-reduce paradigm, with mappers and reducers working on multiple machines \n",
      "\n",
      "Then we calculate the k-nearest neighbors in the 'space' of the user: this involves a database lookup and an iteration over the items a user has rated. Since the latter is usually not a very large number, this computation can be managed on a front end machine (even if storing the database will take a lot of memory).\n"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "We'll first create subset data frames, which have just those columns which we will send to the map-reduce. We'll also strip out the header and index of the frame. The reason for doing this is: unless we pre-populate the machines on Amazon with software, we can *rely only on the regular python library, numpy, and scipy being there (and at python 2.6)*, and thus we will need to parse the csv file, line by line (`mrjob` uses hadoop's stream protocol and thus needs to be fed line by line)."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "subsetoffull=fulldf[['user_id','business_id', 'stars','business_avg','user_avg']]\n",
      "subsetoffull.to_csv(\"subset-full.csv\", index=False, header=False)\n",
      "subsetofsmall=smalldf[['user_id','business_id', 'stars','business_avg','user_avg']]\n",
      "subsetofsmall.to_csv(\"subset-small.csv\", index=False, header=False)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 46
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "### Running mrjob locally\n",
      "\n",
      "`mrjob` scripts cannot be run from the ipython notebook, as they fork themselves on execution. Thus you must write the code for mrjob in a separate file which you must submit along with this homework, in the same folder as the python notebook file.\n",
      "\n",
      "If you have not done so already (you were supposed to do this as part of HW 0), you will first need to install `mrjob`. The appropriate equivalent of the following incantation should do the job:\n",
      "\n",
      "    ~/anaconda/bin/pip install mrjob\n",
      "    \n",
      "\n",
      "    \n",
      "To familiarize yourself with the structure of an `mrjob` script, please read [this](http://mrjob.readthedocs.org/en/latest/guides/quickstart.html#writing-your-first-job) . Run the examples in that document to familiarize yourself with `mrjob`.\n",
      "\n",
      "The kind of script you will be writing is in the section \"Writing your second job\" in that document. \n"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "All mrjob tasks use the map-reduce strategy to divide up computation across computers. You should work through the mrjob tutorial to gain familiarity with this, but we\u2019ll also outline the basic process here:\n",
      "\n",
      "1. During the first map step, mrjob calls a mapper function with a key (which for the first step is None), and a value (which for the first step is a line of data from an input file). This function does whatever it wants with this data, and yields a key and value.  The key is used in step 2 to gather up the values from all the different mappers into groups\n",
      "\n",
      "2. mrjob collects the outputs from all the mappers, and gathers them into subsets with the same key value (this is similar to what pandas.groupby does). It passes each of these subsets to a reducer (or \u201ccollector\u201d) function, whose job is to synthesize this list of grouped data into something useful (e.g., computing the mean of all the inputs). It then yields the key and reduced value. \n",
      "\n",
      "3. If there are any additional steps, mrjob feeds each output from a reducer function in step 2 to the next mapper. Otherwise, it prints the output.\n",
      "\n",
      "The point behind map-reduce is to agree upon a common framework to split up a large computational job into smaller tasks. mrjob then has a lot of freedom to organize how these tasks run in parallel, on many machines"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "#### Writing your script"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**4.1** Write a MRJOB script, called `computesim.py`. The object of this script is to take a csv file and return a tuple `(rho, n_common)` as `calculate_similarity` for pairs of restaurants. See `skeleton.py` below for the SPEC of this file. Your job is to fill in those methods. You MUST use this skeleton.\n",
      "\n",
      "This script is to be run like so (substitute your own operating system's call):\n",
      "\n",
      "    ~/anaconda/bin/python computesim.py subset-small.csv > output.small.local.txt\n",
      "\n",
      "Thus, when the script below is run in this fashion, mrjob will read the data line-by-line from subset-small.csv, and pass it to the first \"step\".\n",
      "\n"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "####Algorithm to calculate pearson similarities\n",
      "\n",
      "Here is the description of the algorithm for RestaurantSimilarities.\n",
      "\n",
      "Your code will have two steps. Each step will have a mapper and a reducer. These are described in turn here:\n",
      "\n",
      "1. `line_mapper` will split the line, yielding the `user_id` as key, and the rest as value. This method's implementation is provided for you.\n",
      "\n",
      "2. `users_items_collector` is a reducer. It is passed ALL mapper outputs corresponding to a particular `user_id`. Put these emissions into a list, and re-emit the `user_id` with this list.\n",
      "\n",
      "3. `pair_items_mapper` takes the `user_id` and the list. It dosent do anything with the `user_id`, however, it takes every combination (thus len(list) choose 2) of 2 `business_id`s from the passed on list (see combinations in itertools in the python documentation) and sends on the remaining information keyed on the tuple `(restaurant1, restaurant2)`. Be sure to handle the case where the restaurant id's are flipped: include them somehow under the same key.\n",
      "\n",
      "4. `calc_sim_collector` is passed ALL sent on list information for the pair of restaurants that was emitted in the previous step. Note that thse will come from different `user_id`s. This sort of collection is key to this style of programming. This list information should now correspond to all the common support of the two restaurants. Use this information to calculate this common support and the pearson similarity. Return the aforementioned tuple by yielding it keyed by the tuple of restaurants. This information will be sent to the output file. The output keys and values will both be in JSON format, separated by a tab.\n",
      "\n",
      "The output should be saved in a file via redirection as `output.small.local.txt`"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "####Skeleton File for this problem\n",
      "\n",
      "You can access it [here](https://raw.github.com/cs109/content/master/skeleton.py) or just run the next cell to see it."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from pygments import highlight\n",
      "from pygments.lexers import PythonLexer\n",
      "from pygments.formatters import HtmlFormatter\n",
      "from IPython.display import HTML\n",
      "import urllib\n",
      "skelcode = urllib.urlopen(\"https://raw.github.com/cs109/content/master/skeleton.py\").read()\n",
      "skelhtml=highlight(skelcode, PythonLexer(), HtmlFormatter())\n",
      "HTML(skelhtml)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "html": [
        "<div class=\"highlight\"><pre><span class=\"kn\">import</span> <span class=\"nn\">numpy</span> <span class=\"kn\">as</span> <span class=\"nn\">np</span>\n",
        "\n",
        "<span class=\"kn\">from</span> <span class=\"nn\">mrjob.job</span> <span class=\"kn\">import</span> <span class=\"n\">MRJob</span>\n",
        "<span class=\"kn\">from</span> <span class=\"nn\">itertools</span> <span class=\"kn\">import</span> <span class=\"n\">combinations</span><span class=\"p\">,</span> <span class=\"n\">permutations</span>\n",
        "\n",
        "<span class=\"kn\">from</span> <span class=\"nn\">scipy.stats.stats</span> <span class=\"kn\">import</span> <span class=\"n\">pearsonr</span>\n",
        "\n",
        "\n",
        "<span class=\"k\">class</span> <span class=\"nc\">RestaurantSimilarities</span><span class=\"p\">(</span><span class=\"n\">MRJob</span><span class=\"p\">):</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">steps</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">):</span>\n",
        "        <span class=\"s\">&quot;the steps in the map-reduce process&quot;</span>\n",
        "        <span class=\"n\">thesteps</span> <span class=\"o\">=</span> <span class=\"p\">[</span>\n",
        "            <span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">mr</span><span class=\"p\">(</span><span class=\"n\">mapper</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">line_mapper</span><span class=\"p\">,</span> <span class=\"n\">reducer</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">users_items_collector</span><span class=\"p\">),</span>\n",
        "            <span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">mr</span><span class=\"p\">(</span><span class=\"n\">mapper</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">pair_items_mapper</span><span class=\"p\">,</span> <span class=\"n\">reducer</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">calc_sim_collector</span><span class=\"p\">)</span>\n",
        "        <span class=\"p\">]</span>\n",
        "        <span class=\"k\">return</span> <span class=\"n\">thesteps</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">line_mapper</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span><span class=\"n\">_</span><span class=\"p\">,</span><span class=\"n\">line</span><span class=\"p\">):</span>\n",
        "        <span class=\"s\">&quot;this is the complete implementation&quot;</span>\n",
        "        <span class=\"n\">user_id</span><span class=\"p\">,</span><span class=\"n\">business_id</span><span class=\"p\">,</span><span class=\"n\">stars</span><span class=\"p\">,</span><span class=\"n\">business_avg</span><span class=\"p\">,</span><span class=\"n\">user_avg</span><span class=\"o\">=</span><span class=\"n\">line</span><span class=\"o\">.</span><span class=\"n\">split</span><span class=\"p\">(</span><span class=\"s\">&#39;,&#39;</span><span class=\"p\">)</span>\n",
        "        <span class=\"k\">yield</span> <span class=\"n\">user_id</span><span class=\"p\">,</span> <span class=\"p\">(</span><span class=\"n\">business_id</span><span class=\"p\">,</span><span class=\"n\">stars</span><span class=\"p\">,</span><span class=\"n\">business_avg</span><span class=\"p\">,</span><span class=\"n\">user_avg</span><span class=\"p\">)</span>\n",
        "\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">users_items_collector</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span> <span class=\"n\">user_id</span><span class=\"p\">,</span> <span class=\"n\">values</span><span class=\"p\">):</span>\n",
        "        <span class=\"sd\">&quot;&quot;&quot;</span>\n",
        "<span class=\"sd\">        #iterate over the list of tuples yielded in the previous mapper</span>\n",
        "<span class=\"sd\">        #and append them to an array of rating information</span>\n",
        "<span class=\"sd\">        &quot;&quot;&quot;</span>\n",
        "        <span class=\"k\">pass</span>\n",
        "\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">pair_items_mapper</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span> <span class=\"n\">user_id</span><span class=\"p\">,</span> <span class=\"n\">values</span><span class=\"p\">):</span>\n",
        "        <span class=\"sd\">&quot;&quot;&quot;</span>\n",
        "<span class=\"sd\">        ignoring the user_id key, take all combinations of business pairs</span>\n",
        "<span class=\"sd\">        and yield as key the pair id, and as value the pair rating information</span>\n",
        "<span class=\"sd\">        &quot;&quot;&quot;</span>\n",
        "\t   <span class=\"k\">pass</span> <span class=\"c\">#your code here</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">calc_sim_collector</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span> <span class=\"n\">key</span><span class=\"p\">,</span> <span class=\"n\">values</span><span class=\"p\">):</span>\n",
        "        <span class=\"sd\">&quot;&quot;&quot;</span>\n",
        "<span class=\"sd\">        Pick up the information from the previous yield as shown. Compute</span>\n",
        "<span class=\"sd\">        the pearson correlation and yield the final information as in the</span>\n",
        "<span class=\"sd\">        last line here.</span>\n",
        "<span class=\"sd\">        &quot;&quot;&quot;</span>\n",
        "        <span class=\"p\">(</span><span class=\"n\">rest1</span><span class=\"p\">,</span> <span class=\"n\">rest2</span><span class=\"p\">),</span> <span class=\"n\">common_ratings</span> <span class=\"o\">=</span> <span class=\"n\">key</span><span class=\"p\">,</span> <span class=\"n\">values</span>\n",
        "\t    <span class=\"c\">#your code here</span>\n",
        "        <span class=\"k\">yield</span> <span class=\"p\">(</span><span class=\"n\">rest1</span><span class=\"p\">,</span> <span class=\"n\">rest2</span><span class=\"p\">),</span> <span class=\"p\">(</span><span class=\"n\">rho</span><span class=\"p\">,</span> <span class=\"n\">n_common</span><span class=\"p\">)</span>\n",
        "\n",
        "\n",
        "<span class=\"c\">#Below MUST be there for things to work</span>\n",
        "<span class=\"k\">if</span> <span class=\"n\">__name__</span> <span class=\"o\">==</span> <span class=\"s\">&#39;__main__&#39;</span><span class=\"p\">:</span>\n",
        "    <span class=\"n\">RestaurantSimilarities</span><span class=\"o\">.</span><span class=\"n\">run</span><span class=\"p\">()</span>\n",
        "</pre></div>\n"
       ],
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 47,
       "text": [
        "<IPython.core.display.HTML at 0x10ce456d0>"
       ]
      }
     ],
     "prompt_number": 47
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "#### Explanation for those funny `yield` keywords\n",
      "\n",
      "The functions above \u201cyield\u201d values, and do not \u201creturn\u201d them. They are **generators**. Here is an example:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def upper_generator(words):\n",
      "    for word in words:\n",
      "        yield word.upper()\n",
      "\n",
      "words = ['a', 'couple', 'of', 'words', 'to', 'process']\n",
      "\n",
      "print upper_generator(words)\n",
      "print list(upper_generator(words))\n",
      "for u in upper_generator(words):\n",
      "     print u\n",
      "\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "<generator object upper_generator at 0x110b388c0>\n",
        "['A', 'COUPLE', 'OF', 'WORDS', 'TO', 'PROCESS']\n",
        "A\n",
        "COUPLE\n",
        "OF\n",
        "WORDS\n",
        "TO\n",
        "PROCESS\n"
       ]
      }
     ],
     "prompt_number": 48
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "You can read more [here](http://nedbatchelder.com/text/iter.html). Also see Thu Oct 17th's class video for information about classes and generators."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Include `computesim.py` in your submission in the same folder as the notebook. Uncommenting and running the following cell should **output your code in here**."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "thecode = open(\"computesim.py\").read()\n",
      "thehtml=highlight(thecode, PythonLexer(), HtmlFormatter())\n",
      "HTML(thehtml)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "html": [
        "<div class=\"highlight\"><pre><span class=\"kn\">import</span> <span class=\"nn\">numpy</span> <span class=\"kn\">as</span> <span class=\"nn\">np</span>\n",
        "\n",
        "<span class=\"kn\">from</span> <span class=\"nn\">mrjob.job</span> <span class=\"kn\">import</span> <span class=\"n\">MRJob</span>\n",
        "<span class=\"kn\">from</span> <span class=\"nn\">itertools</span> <span class=\"kn\">import</span> <span class=\"n\">combinations</span><span class=\"p\">,</span> <span class=\"n\">permutations</span>\n",
        "<span class=\"kn\">from</span> <span class=\"nn\">math</span> <span class=\"kn\">import</span> <span class=\"n\">sqrt</span>\n",
        "\n",
        "<span class=\"kn\">from</span> <span class=\"nn\">scipy.stats.stats</span> <span class=\"kn\">import</span> <span class=\"n\">pearsonr</span>\n",
        "\n",
        "<span class=\"k\">class</span> <span class=\"nc\">RestaurantSimilarities</span><span class=\"p\">(</span><span class=\"n\">MRJob</span><span class=\"p\">):</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">steps</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">):</span>\n",
        "        <span class=\"n\">thesteps</span> <span class=\"o\">=</span> <span class=\"p\">[</span>\n",
        "            <span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">mr</span><span class=\"p\">(</span><span class=\"n\">mapper</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">line_mapper</span><span class=\"p\">,</span> <span class=\"n\">reducer</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">users_items_collector</span><span class=\"p\">),</span>\n",
        "            <span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">mr</span><span class=\"p\">(</span><span class=\"n\">mapper</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">pair_items_mapper</span><span class=\"p\">,</span> <span class=\"n\">reducer</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">calc_sim_collector</span><span class=\"p\">)</span>\n",
        "        <span class=\"p\">]</span>\n",
        "        <span class=\"k\">return</span> <span class=\"n\">thesteps</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">line_mapper</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span><span class=\"n\">_</span><span class=\"p\">,</span><span class=\"n\">line</span><span class=\"p\">):</span>\n",
        "        <span class=\"n\">user_id</span><span class=\"p\">,</span><span class=\"n\">business_id</span><span class=\"p\">,</span><span class=\"n\">stars</span><span class=\"p\">,</span><span class=\"n\">business_avg</span><span class=\"p\">,</span><span class=\"n\">user_avg</span><span class=\"o\">=</span><span class=\"n\">line</span><span class=\"o\">.</span><span class=\"n\">split</span><span class=\"p\">(</span><span class=\"s\">&#39;,&#39;</span><span class=\"p\">)</span>\n",
        "        <span class=\"k\">yield</span> <span class=\"n\">user_id</span><span class=\"p\">,</span> <span class=\"p\">(</span><span class=\"n\">business_id</span><span class=\"p\">,</span><span class=\"n\">stars</span><span class=\"p\">,</span><span class=\"n\">business_avg</span><span class=\"p\">,</span><span class=\"n\">user_avg</span><span class=\"p\">)</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">users_items_collector</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span> <span class=\"n\">user_id</span><span class=\"p\">,</span> <span class=\"n\">values</span><span class=\"p\">):</span>\n",
        "        <span class=\"n\">ratings</span><span class=\"o\">=</span><span class=\"p\">[]</span>\n",
        "        <span class=\"k\">for</span> <span class=\"n\">business_id</span><span class=\"p\">,</span><span class=\"n\">stars</span><span class=\"p\">,</span><span class=\"n\">business_avg</span><span class=\"p\">,</span><span class=\"n\">user_avg</span> <span class=\"ow\">in</span> <span class=\"n\">values</span><span class=\"p\">:</span>\n",
        "            <span class=\"n\">ratings</span><span class=\"o\">.</span><span class=\"n\">append</span><span class=\"p\">((</span><span class=\"n\">business_id</span><span class=\"p\">,(</span><span class=\"n\">stars</span><span class=\"p\">,</span> <span class=\"n\">user_avg</span><span class=\"p\">)))</span>\n",
        "        <span class=\"k\">yield</span> <span class=\"n\">user_id</span><span class=\"p\">,</span> <span class=\"n\">ratings</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">pair_items_mapper</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span> <span class=\"n\">user_id</span><span class=\"p\">,</span> <span class=\"n\">values</span><span class=\"p\">):</span>\n",
        "        <span class=\"n\">ratings</span> <span class=\"o\">=</span> <span class=\"n\">values</span>\n",
        "        <span class=\"k\">for</span> <span class=\"n\">biz1tuple</span><span class=\"p\">,</span> <span class=\"n\">biz2tuple</span> <span class=\"ow\">in</span> <span class=\"n\">combinations</span><span class=\"p\">(</span><span class=\"n\">ratings</span><span class=\"p\">,</span> <span class=\"mi\">2</span><span class=\"p\">):</span>\n",
        "            <span class=\"n\">biz1</span><span class=\"p\">,</span> <span class=\"n\">biz1r</span><span class=\"o\">=</span><span class=\"n\">biz1tuple</span>\n",
        "            <span class=\"n\">biz2</span><span class=\"p\">,</span> <span class=\"n\">biz2r</span><span class=\"o\">=</span><span class=\"n\">biz2tuple</span>\n",
        "            <span class=\"k\">if</span> <span class=\"n\">biz1</span> <span class=\"o\">&lt;=</span> <span class=\"n\">biz2</span> <span class=\"p\">:</span>\n",
        "                <span class=\"k\">yield</span> <span class=\"p\">(</span><span class=\"n\">biz1</span><span class=\"p\">,</span> <span class=\"n\">biz2</span><span class=\"p\">),</span> <span class=\"p\">(</span><span class=\"n\">biz1r</span><span class=\"p\">,</span> <span class=\"n\">biz2r</span><span class=\"p\">)</span>\n",
        "            <span class=\"k\">else</span><span class=\"p\">:</span>\n",
        "                <span class=\"k\">yield</span> <span class=\"p\">(</span><span class=\"n\">biz2</span><span class=\"p\">,</span> <span class=\"n\">biz1</span><span class=\"p\">),</span> <span class=\"p\">(</span><span class=\"n\">biz2r</span><span class=\"p\">,</span> <span class=\"n\">biz1r</span><span class=\"p\">)</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">calc_sim_collector</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span> <span class=\"n\">key</span><span class=\"p\">,</span> <span class=\"n\">values</span><span class=\"p\">):</span>\n",
        "        <span class=\"p\">(</span><span class=\"n\">rest1</span><span class=\"p\">,</span> <span class=\"n\">rest2</span><span class=\"p\">),</span> <span class=\"n\">common_ratings</span> <span class=\"o\">=</span> <span class=\"n\">key</span><span class=\"p\">,</span> <span class=\"n\">values</span>\n",
        "        <span class=\"n\">diff1</span><span class=\"o\">=</span><span class=\"p\">[]</span>\n",
        "        <span class=\"n\">diff2</span><span class=\"o\">=</span><span class=\"p\">[]</span>\n",
        "        <span class=\"n\">n_common</span><span class=\"o\">=</span><span class=\"mi\">0</span>\n",
        "\n",
        "\n",
        "        <span class=\"k\">for</span> <span class=\"n\">rt1</span><span class=\"p\">,</span> <span class=\"n\">rt2</span> <span class=\"ow\">in</span> <span class=\"n\">common_ratings</span><span class=\"p\">:</span>\n",
        "            <span class=\"n\">diff1</span><span class=\"o\">.</span><span class=\"n\">append</span><span class=\"p\">(</span><span class=\"nb\">float</span><span class=\"p\">(</span><span class=\"n\">rt1</span><span class=\"p\">[</span><span class=\"mi\">0</span><span class=\"p\">])</span><span class=\"o\">-</span><span class=\"nb\">float</span><span class=\"p\">(</span><span class=\"n\">rt1</span><span class=\"p\">[</span><span class=\"mi\">1</span><span class=\"p\">]))</span>\n",
        "            <span class=\"n\">diff2</span><span class=\"o\">.</span><span class=\"n\">append</span><span class=\"p\">(</span><span class=\"nb\">float</span><span class=\"p\">(</span><span class=\"n\">rt2</span><span class=\"p\">[</span><span class=\"mi\">0</span><span class=\"p\">])</span><span class=\"o\">-</span><span class=\"nb\">float</span><span class=\"p\">(</span><span class=\"n\">rt2</span><span class=\"p\">[</span><span class=\"mi\">1</span><span class=\"p\">]))</span>\n",
        "            <span class=\"n\">n_common</span><span class=\"o\">=</span><span class=\"n\">n_common</span><span class=\"o\">+</span><span class=\"mi\">1</span>\n",
        "        <span class=\"k\">if</span> <span class=\"n\">n_common</span><span class=\"o\">==</span><span class=\"mi\">0</span><span class=\"p\">:</span>\n",
        "            <span class=\"n\">rho</span><span class=\"o\">=</span><span class=\"mf\">0.</span>\n",
        "        <span class=\"k\">else</span><span class=\"p\">:</span>\n",
        "            <span class=\"n\">rho</span><span class=\"o\">=</span><span class=\"n\">pearsonr</span><span class=\"p\">(</span><span class=\"n\">diff1</span><span class=\"p\">,</span> <span class=\"n\">diff2</span><span class=\"p\">)[</span><span class=\"mi\">0</span><span class=\"p\">]</span>\n",
        "            <span class=\"k\">if</span> <span class=\"n\">np</span><span class=\"o\">.</span><span class=\"n\">isnan</span><span class=\"p\">(</span><span class=\"n\">rho</span><span class=\"p\">):</span>\n",
        "                <span class=\"n\">rho</span><span class=\"o\">=</span><span class=\"mf\">0.</span>\n",
        "        <span class=\"k\">yield</span> <span class=\"p\">(</span><span class=\"n\">rest1</span><span class=\"p\">,</span> <span class=\"n\">rest2</span><span class=\"p\">),</span> <span class=\"p\">(</span><span class=\"n\">rho</span><span class=\"p\">,</span> <span class=\"n\">n_common</span><span class=\"p\">)</span>\n",
        "\n",
        "\n",
        "<span class=\"c\">#Below MUST be there for things to work!</span>\n",
        "<span class=\"k\">if</span> <span class=\"n\">__name__</span> <span class=\"o\">==</span> <span class=\"s\">&#39;__main__&#39;</span><span class=\"p\">:</span>\n",
        "    <span class=\"n\">RestaurantSimilarities</span><span class=\"o\">.</span><span class=\"n\">run</span><span class=\"p\">()</span>\n",
        "</pre></div>\n"
       ],
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 49,
       "text": [
        "<IPython.core.display.HTML at 0x10cef1f10>"
       ]
      }
     ],
     "prompt_number": 49
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "#### Checking the results\n",
      "\n",
      "Let us load the data from the file"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "output_small_local=[[json.loads(j) for j in line.strip().split(\"\\t\")] for line in open(\"./output.small.local.txt\")]\n",
      "output_small_local[0]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 50,
       "text": [
        "[[u'-4A5xmN21zi_TXnUESauUQ', u'-AAig9FG0s8gYE4f8GfowQ'],\n",
        " [0.384365693729571, 5]]"
       ]
      }
     ],
     "prompt_number": 50
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "We will Implement a function `make_database_from_pairs` which takes a dataframe of restaurants `smalldf` and the output parsed in the previous command to create the database like before. By the nature of the map-reduce algorithms these only contain those restaurant pairs with common support. The `Database` constructor initializes the remaining similarities to 0.\n",
      "\n",
      "The function will take the dataframe and `bizpairs` obtained by parsing the EMR output file which have the key of business pairs and value the pair of pearson correlation and `n_common`. It will return an instance of the `Database` class.\n",
      "\n",
      "This function will take a long time to run on large data sets.\n"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def make_database_from_pairs(df, bizpairs):\n",
      "    \"\"\"\n",
      "    make the database from the pairs returned from mrjob.\n",
      "    df is the dataframe, smalldf or fulldf.\n",
      "    bizpairs are a list of elements, each of which is a list of two\n",
      "        lists. The first of these lists has the two business id's, while\n",
      "        the second has the similarity and the common support\n",
      "    Returns an instance of the Database class.\n",
      "    \"\"\"\n",
      "    dbase=Database(df)\n",
      "    cache={}\n",
      "    for bp,corrs in bizpairs:\n",
      "        b1,b2=bp\n",
      "        i1=dbase.uniquebizids[b1]\n",
      "        i2=dbase.uniquebizids[b2]\n",
      "        sim,nsup=corrs\n",
      "        dbase.database_sim[i1][i2]=sim\n",
      "        dbase.database_sim[i2][i1]=sim\n",
      "        dbase.database_sup[i1][i2]=nsup\n",
      "        dbase.database_sup[i2][i1]=nsup\n",
      "        if cache.has_key(b1):\n",
      "            nsup1=cache[b1]\n",
      "        else:\n",
      "            nsup1=dbase.df[dbase.df.business_id==b1].user_id.count()\n",
      "            cache[b1]=nsup1\n",
      "        if cache.has_key(b2):\n",
      "            nsup2=cache[b2]\n",
      "        else:\n",
      "            nsup2=dbase.df[dbase.df.business_id==b2].user_id.count()\n",
      "            cache[b2]=nsup2\n",
      "        dbase.database_sim[i1][i1]=1.0\n",
      "        dbase.database_sim[i2][i2]=1.0\n",
      "        dbase.database_sup[i1][i1]=nsup1\n",
      "        dbase.database_sup[i2][i2]=nsup2\n",
      "    return dbase"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 51
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "We will store the output in variable `db_mrjob_local`."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "db_mrjob_local=make_database_from_pairs(smalldf, output_small_local)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 52
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "We print a pair to see that our answers are identical."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print db.get(\"zruUQvFySeXyEd7_rQixBg\", \"z3yFuLVrmH-3RJruPEMYKw\")\n",
      "print db_mrjob_local.get(\"zruUQvFySeXyEd7_rQixBg\", \"z3yFuLVrmH-3RJruPEMYKw\")"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "(0.39904554525734559, 7)\n",
        "(0.39904554525734542, 7)\n"
       ]
      }
     ],
     "prompt_number": 53
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**4.2** Lets test that our results are overall the same as before"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "sums=0.\n",
      "count=0\n",
      "for k in db.uniquebizids.keys():\n",
      "    for k2 in db.uniquebizids.keys():\n",
      "        count=count+1\n",
      "        sums=sums+db.get(k,k2)[0]-db_mrjob_local.get(k,k2)[0]\n",
      "print sums, count"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "-8.65973959208e-15 29584\n"
       ]
      }
     ],
     "prompt_number": 54
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "### Running on Amazon Elastic Map Reduce(EMR)"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "At this point, we shall shift to running on Amazon EMR. \n",
      "\n",
      "------------\n",
      "\n",
      "*Read [this document](http://nbviewer.ipython.org/urls/raw.github.com/cs109/content/master/InstructionsForAmazonEMR.ipynb) for instructions on how to set yourself up on Amazon.*\n",
      "\n",
      "------------\n",
      "\n",
      "#### Reproduce the results with the smaller file on EMR\n",
      "\n",
      "Test the smaller file and make sure it has the same results. For example, you could use the incantation:\n",
      "\n",
      "    ~/anaconda/bin/python computesim.py -r emr --num-ec2-instances 2 subset-small.csv > output.small.emr.txt\n",
      "\n",
      "You do **NOT** need to submit any results from that exploration to us.\n",
      "\n",
      "**Important**: Please always make sure that your code is bug free, before actually submitting it to amazon. Try to run the job locally first and see if it produces the desired result. Then, if this worked, you are ready to proceed to the cloud. The homework problems are small and your free credit should provide you with a lot of room for running and testing on Amazon. However, it is your responsibility to make sure the jobs terminate properly and do not cause excessive costs.\n",
      "\n",
      "You can always monitor your currently running jobs (in the US-East sector) using [this overview at region US-EAST-1](https://console.aws.amazon.com/elasticmapreduce/home?region=us-east-1) of your MapReduce job flows."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "#### Running the larger job"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**4.3** Run the script on the larger file `subset-full.csv`. Use between 4-8 instances on EMR on Amazon. Save the output in `output.full.emr.txt`. Your incantation will be something like:\n",
      "\n",
      "    ~/anaconda/bin/python computesim.py -r emr --num-ec2-instances 5 subset-full.csv > output.full.emr.txt"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "You might elect to [save the file on S3](http://mrjob.readthedocs.org/en/latest/guides/emr-quickstart.html#sending-output-to-a-specific-place) and bring it over manually.\n",
      "\n",
      "Try and think about what size job would be best to run on Amazon, given that there is a setup time. There is a way to persistently set up machines (the mrjob documentation provides the details), but then remember you will be billed for that setup and need to monitor it. However, a persistent setup might come useful for your projects."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Loading the full output from EMR"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Lets load the output in. **CAUTION** The next two cells will also take a lot of time to run and load.  "
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "output_full_emr=[[json.loads(j) for j in l.strip().split(\"\\t\")] for l in open(\"./output.full.emr.txt\")]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 55
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "This function will take a very long time to run, on the order of 5 minutes or more, depending on your computer"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "dbfull=make_database_from_pairs(fulldf, output_full_emr)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 56
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**4.4**  For `testuserid`, once again, print out the ratings using the `bizs` list as before. How have they changed with respect to Question 2? Why might this be?"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "#your code here\n",
      "print \"for user\",usernamefromid(fulldf,testuserid), 'avg', fulldf[fulldf.user_id==testuserid].stars.mean() \n",
      "for i in bizs:\n",
      "    print \"=========\"\n",
      "    print biznamefromid(fulldf, i), i\n",
      "    print rating(fulldf, dbfull, i, testuserid, k=7, reg=3.) \n",
      "    u,a=get_other_ratings(i, testuserid, fulldf)\n",
      "    print \"User Score:\",u,\"Avg score\",a"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "for user Vern avg 3.58227848101\n",
        "=========\n",
        "Local Breeze"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " GAPqG0WNBBidKeZTMpEZ-w\n",
        "4.81062901451"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "User Score: 5 Avg score 3.86363636364\n",
        "=========\n",
        "Carly's Bistro zmFc8M-hS4uuyY0hklIpoQ\n",
        "4.79797424984"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "User Score: 5 Avg score 3.65079365079\n",
        "=========\n",
        "Tee Pee Mexican Food"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " soiGohHtWOltGeomkSxzEw\n",
        "4.23927633965"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "User Score: 5 Avg score 3.13636363636\n",
        "=========\n",
        "District American Kitchen and Wine Bar 9ziO3NpoNTKHvIKCBFB_fQ\n",
        "4.08622622369"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "User Score: 4 Avg score 3.575\n",
        "=========\n",
        "Los Reyes de la Torta bzDs0u8I-z231QVdIQWkrA\n",
        "3.94718985123"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "User Score: 4 Avg score 4.24796747967\n"
       ]
      }
     ],
     "prompt_number": 57
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*your answer here*\n",
      "\n",
      "I copy here the results from the small set:\n",
      "\n",
      "    for user Vern avg 3.5652173913\n",
      "    ----------------------------------\n",
      "    Local Breeze\n",
      "    Predicted Rating: 4.2280987611\n",
      "    Actual User Rating: 5 Avg Rating 4.0\n",
      "    ----------------------------------\n",
      "    Carly's Bistro\n",
      "    Predicted Rating: 3.99008654065\n",
      "    Actual User Rating: 5 Avg Rating 3.5\n",
      "    ----------------------------------\n",
      "    Tee Pee Mexican Food\n",
      "    Predicted Rating: 3.52640184162\n",
      "    Actual User Rating: 5 Avg Rating 3.04347826087\n",
      "    ----------------------------------\n",
      "    District American Kitchen and Wine Bar\n",
      "    Predicted Rating: 3.80281696528\n",
      "    Actual User Rating: 4 Avg Rating 3.55263157895\n",
      "    ----------------------------------\n",
      "    Los Reyes de la Torta\n",
      "    Predicted Rating: 3.41514298977\n",
      "    Actual User Rating: 4 Avg Rating 4.13157894737\n",
      "    \n",
      "One can see that the predicted rating is much closer to the actual one in the larger set. The reason for this:\n",
      "We are still working with user Vern from the previous set, so we are choosing a \"prolific\" user who is likely to have a dense neighborhood of restaurants in the larger set (by the choice of our cut to create the small set).\n",
      "\n",
      "Note that if we evalusted our predictions on the entire data set, rather than cherry-picking users like Vern, our average prediction error may become worse. There is because the larger data set is more likely to be of higher sparsity, as we will include more users who rated very few restaurants."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "**4.5** Outline another `step` (in words) in the mrjob map-reduce class to implement a simple but scalable recommender of the global type that we did in Question 1.5 to 1.7."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*your answer here*\n",
      "\n",
      "We can add one additional map and one additional reduce step to achieve 1.5 and 1.6.\n",
      "\n",
      "MAP STEP:\n",
      "    \n",
      "    def ranking_mapper(self, restaurants, values):\n",
      "            sim, n_common = values\n",
      "            rest1, rest2 = restaurants\n",
      "            if int(n_common) > 0:\n",
      "                yield (rest1, sim), (rest2, n_common)\n",
      "\n",
      "\n",
      "REDUCE STEP:\n",
      "        \n",
      "        def top_similar_collector(self, key, values):\n",
      "            rest1, sim = key\n",
      "            for rest2, n_common in values:\n",
      "                yield None, (rest1, rest2, sim, n_common)\n",
      "                \n",
      "Note that by default mrjob does an alphanumeric sort on sim, which is not what we want. Indeed it is complicated, but possible to have this addition run locally (on mac/linux) you would need to specify the sort binary as `sort -nr`. On Hadoop, the parameters outlined [here](http://pythonhosted.org/mrjob/job.html#mrjob.job.MRJob.jobconf) can be used. Of-course, you could do the final sorting on a front end machine anyway.\n",
      "\n",
      "UPDATE: Brandon suggests an even simpler solution that will work both locally and on EMR, without having to use any hadoop sorting specifics. \n",
      "We change to:\n",
      "\n",
      "MAP STEP:\n",
      "    \n",
      "    def ranking_mapper(self, restaurants, values):\n",
      "            sim, n_common = values\n",
      "            rest1, rest2 = restaurants\n",
      "            if int(n_common) > 0:\n",
      "                yield (rest1), (sim, rest2, n_common)\n",
      "\n",
      "\n",
      "REDUCE STEP:\n",
      "        \n",
      "        def top_similar_collector(self, key, values):\n",
      "            rest1 = key\n",
      "            for sim, rest2, n_common in sorted(values, reverse=True):\n",
      "                yield None, (rest1, rest2, sim, n_common)\n",
      "                \n",
      "Full code:\n"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "thecode = open(\"computesim2.py\").read()\n",
      "thehtml=highlight(thecode, PythonLexer(), HtmlFormatter())\n",
      "HTML(thehtml)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "html": [
        "<div class=\"highlight\"><pre><span class=\"kn\">import</span> <span class=\"nn\">numpy</span> <span class=\"kn\">as</span> <span class=\"nn\">np</span>\n",
        "\n",
        "<span class=\"kn\">from</span> <span class=\"nn\">mrjob.job</span> <span class=\"kn\">import</span> <span class=\"n\">MRJob</span>\n",
        "<span class=\"kn\">from</span> <span class=\"nn\">itertools</span> <span class=\"kn\">import</span> <span class=\"n\">combinations</span><span class=\"p\">,</span> <span class=\"n\">permutations</span>\n",
        "<span class=\"kn\">from</span> <span class=\"nn\">math</span> <span class=\"kn\">import</span> <span class=\"n\">sqrt</span>\n",
        "<span class=\"kn\">import</span> <span class=\"nn\">mrjob</span>\n",
        "\n",
        "<span class=\"kn\">from</span> <span class=\"nn\">scipy.stats.stats</span> <span class=\"kn\">import</span> <span class=\"n\">pearsonr</span>\n",
        "\n",
        "<span class=\"k\">class</span> <span class=\"nc\">RestaurantSimilarities</span><span class=\"p\">(</span><span class=\"n\">MRJob</span><span class=\"p\">):</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">steps</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">):</span>\n",
        "        <span class=\"n\">thesteps</span> <span class=\"o\">=</span> <span class=\"p\">[</span>\n",
        "            <span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">mr</span><span class=\"p\">(</span><span class=\"n\">mapper</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">line_mapper</span><span class=\"p\">,</span> <span class=\"n\">reducer</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">users_items_collector</span><span class=\"p\">),</span>\n",
        "            <span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">mr</span><span class=\"p\">(</span><span class=\"n\">mapper</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">pair_items_mapper</span><span class=\"p\">,</span> <span class=\"n\">reducer</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">calc_sim_collector</span><span class=\"p\">),</span>\n",
        "            <span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">mr</span><span class=\"p\">(</span><span class=\"n\">mapper</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">ranking_mapper</span><span class=\"p\">,</span> <span class=\"n\">reducer</span><span class=\"o\">=</span><span class=\"bp\">self</span><span class=\"o\">.</span><span class=\"n\">top_similar_collector</span><span class=\"p\">)</span>\n",
        "        <span class=\"p\">]</span>\n",
        "        <span class=\"k\">return</span> <span class=\"n\">thesteps</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">line_mapper</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span><span class=\"n\">_</span><span class=\"p\">,</span><span class=\"n\">line</span><span class=\"p\">):</span>\n",
        "        <span class=\"n\">user_id</span><span class=\"p\">,</span><span class=\"n\">business_id</span><span class=\"p\">,</span><span class=\"n\">stars</span><span class=\"p\">,</span><span class=\"n\">business_avg</span><span class=\"p\">,</span><span class=\"n\">user_avg</span><span class=\"o\">=</span><span class=\"n\">line</span><span class=\"o\">.</span><span class=\"n\">split</span><span class=\"p\">(</span><span class=\"s\">&#39;,&#39;</span><span class=\"p\">)</span>\n",
        "        <span class=\"k\">yield</span> <span class=\"n\">user_id</span><span class=\"p\">,</span> <span class=\"p\">(</span><span class=\"n\">business_id</span><span class=\"p\">,</span><span class=\"n\">stars</span><span class=\"p\">,</span><span class=\"n\">business_avg</span><span class=\"p\">,</span><span class=\"n\">user_avg</span><span class=\"p\">)</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">users_items_collector</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span> <span class=\"n\">user_id</span><span class=\"p\">,</span> <span class=\"n\">values</span><span class=\"p\">):</span>\n",
        "        <span class=\"n\">ratings</span><span class=\"o\">=</span><span class=\"p\">[]</span>\n",
        "        <span class=\"k\">for</span> <span class=\"n\">business_id</span><span class=\"p\">,</span><span class=\"n\">stars</span><span class=\"p\">,</span><span class=\"n\">business_avg</span><span class=\"p\">,</span><span class=\"n\">user_avg</span> <span class=\"ow\">in</span> <span class=\"n\">values</span><span class=\"p\">:</span>\n",
        "            <span class=\"n\">ratings</span><span class=\"o\">.</span><span class=\"n\">append</span><span class=\"p\">((</span><span class=\"n\">business_id</span><span class=\"p\">,(</span><span class=\"n\">stars</span><span class=\"p\">,</span> <span class=\"n\">user_avg</span><span class=\"p\">)))</span>\n",
        "        <span class=\"k\">yield</span> <span class=\"n\">user_id</span><span class=\"p\">,</span> <span class=\"n\">ratings</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">pair_items_mapper</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span> <span class=\"n\">user_id</span><span class=\"p\">,</span> <span class=\"n\">values</span><span class=\"p\">):</span>\n",
        "        <span class=\"n\">ratings</span> <span class=\"o\">=</span> <span class=\"n\">values</span>\n",
        "        <span class=\"k\">for</span> <span class=\"n\">biz1tuple</span><span class=\"p\">,</span> <span class=\"n\">biz2tuple</span> <span class=\"ow\">in</span> <span class=\"n\">combinations</span><span class=\"p\">(</span><span class=\"n\">ratings</span><span class=\"p\">,</span> <span class=\"mi\">2</span><span class=\"p\">):</span>\n",
        "            <span class=\"n\">biz1</span><span class=\"p\">,</span> <span class=\"n\">biz1r</span><span class=\"o\">=</span><span class=\"n\">biz1tuple</span>\n",
        "            <span class=\"n\">biz2</span><span class=\"p\">,</span> <span class=\"n\">biz2r</span><span class=\"o\">=</span><span class=\"n\">biz2tuple</span>\n",
        "            <span class=\"k\">if</span> <span class=\"n\">biz1</span> <span class=\"o\">&lt;=</span> <span class=\"n\">biz2</span> <span class=\"p\">:</span>\n",
        "                <span class=\"k\">yield</span> <span class=\"p\">(</span><span class=\"n\">biz1</span><span class=\"p\">,</span> <span class=\"n\">biz2</span><span class=\"p\">),</span> <span class=\"p\">(</span><span class=\"n\">biz1r</span><span class=\"p\">,</span> <span class=\"n\">biz2r</span><span class=\"p\">)</span>\n",
        "            <span class=\"k\">else</span><span class=\"p\">:</span>\n",
        "                <span class=\"k\">yield</span> <span class=\"p\">(</span><span class=\"n\">biz2</span><span class=\"p\">,</span> <span class=\"n\">biz1</span><span class=\"p\">),</span> <span class=\"p\">(</span><span class=\"n\">biz2r</span><span class=\"p\">,</span> <span class=\"n\">biz1r</span><span class=\"p\">)</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">calc_sim_collector</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span> <span class=\"n\">key</span><span class=\"p\">,</span> <span class=\"n\">values</span><span class=\"p\">):</span>\n",
        "        <span class=\"p\">(</span><span class=\"n\">rest1</span><span class=\"p\">,</span> <span class=\"n\">rest2</span><span class=\"p\">),</span> <span class=\"n\">common_ratings</span> <span class=\"o\">=</span> <span class=\"n\">key</span><span class=\"p\">,</span> <span class=\"n\">values</span>\n",
        "        <span class=\"n\">diff1</span><span class=\"o\">=</span><span class=\"p\">[]</span>\n",
        "        <span class=\"n\">diff2</span><span class=\"o\">=</span><span class=\"p\">[]</span>\n",
        "        <span class=\"n\">n_common</span><span class=\"o\">=</span><span class=\"mi\">0</span>\n",
        "\n",
        "\n",
        "        <span class=\"k\">for</span> <span class=\"n\">rt1</span><span class=\"p\">,</span> <span class=\"n\">rt2</span> <span class=\"ow\">in</span> <span class=\"n\">common_ratings</span><span class=\"p\">:</span>\n",
        "            <span class=\"n\">diff1</span><span class=\"o\">.</span><span class=\"n\">append</span><span class=\"p\">(</span><span class=\"nb\">float</span><span class=\"p\">(</span><span class=\"n\">rt1</span><span class=\"p\">[</span><span class=\"mi\">0</span><span class=\"p\">])</span><span class=\"o\">-</span><span class=\"nb\">float</span><span class=\"p\">(</span><span class=\"n\">rt1</span><span class=\"p\">[</span><span class=\"mi\">1</span><span class=\"p\">]))</span>\n",
        "            <span class=\"n\">diff2</span><span class=\"o\">.</span><span class=\"n\">append</span><span class=\"p\">(</span><span class=\"nb\">float</span><span class=\"p\">(</span><span class=\"n\">rt2</span><span class=\"p\">[</span><span class=\"mi\">0</span><span class=\"p\">])</span><span class=\"o\">-</span><span class=\"nb\">float</span><span class=\"p\">(</span><span class=\"n\">rt2</span><span class=\"p\">[</span><span class=\"mi\">1</span><span class=\"p\">]))</span>\n",
        "            <span class=\"n\">n_common</span><span class=\"o\">=</span><span class=\"n\">n_common</span><span class=\"o\">+</span><span class=\"mi\">1</span>\n",
        "        <span class=\"k\">if</span> <span class=\"n\">n_common</span><span class=\"o\">==</span><span class=\"mi\">0</span><span class=\"p\">:</span>\n",
        "            <span class=\"n\">rho</span><span class=\"o\">=</span><span class=\"mf\">0.</span>\n",
        "        <span class=\"k\">else</span><span class=\"p\">:</span>\n",
        "            <span class=\"n\">rho</span><span class=\"o\">=</span><span class=\"n\">pearsonr</span><span class=\"p\">(</span><span class=\"n\">diff1</span><span class=\"p\">,</span> <span class=\"n\">diff2</span><span class=\"p\">)[</span><span class=\"mi\">0</span><span class=\"p\">]</span>\n",
        "            <span class=\"k\">if</span> <span class=\"n\">np</span><span class=\"o\">.</span><span class=\"n\">isnan</span><span class=\"p\">(</span><span class=\"n\">rho</span><span class=\"p\">):</span>\n",
        "                <span class=\"n\">rho</span><span class=\"o\">=</span><span class=\"mf\">0.</span>\n",
        "        <span class=\"k\">yield</span> <span class=\"p\">(</span><span class=\"n\">rest1</span><span class=\"p\">,</span> <span class=\"n\">rest2</span><span class=\"p\">),</span> <span class=\"p\">(</span><span class=\"n\">rho</span><span class=\"p\">,</span> <span class=\"n\">n_common</span><span class=\"p\">)</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">ranking_mapper</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span> <span class=\"n\">restaurants</span><span class=\"p\">,</span> <span class=\"n\">values</span><span class=\"p\">):</span>\n",
        "        <span class=\"n\">sim</span><span class=\"p\">,</span> <span class=\"n\">n_common</span> <span class=\"o\">=</span> <span class=\"n\">values</span>\n",
        "        <span class=\"n\">rest1</span><span class=\"p\">,</span> <span class=\"n\">rest2</span> <span class=\"o\">=</span> <span class=\"n\">restaurants</span>\n",
        "        <span class=\"k\">if</span> <span class=\"nb\">int</span><span class=\"p\">(</span><span class=\"n\">n_common</span><span class=\"p\">)</span> <span class=\"o\">&gt;</span> <span class=\"mi\">0</span><span class=\"p\">:</span>\n",
        "            <span class=\"k\">yield</span> <span class=\"p\">(</span><span class=\"n\">rest1</span><span class=\"p\">),</span> <span class=\"p\">(</span><span class=\"n\">sim</span><span class=\"p\">,</span> <span class=\"n\">rest2</span><span class=\"p\">,</span> <span class=\"n\">n_common</span><span class=\"p\">)</span>\n",
        "\n",
        "    <span class=\"k\">def</span> <span class=\"nf\">top_similar_collector</span><span class=\"p\">(</span><span class=\"bp\">self</span><span class=\"p\">,</span> <span class=\"n\">key</span><span class=\"p\">,</span> <span class=\"n\">values</span><span class=\"p\">):</span>\n",
        "        <span class=\"n\">rest1</span> <span class=\"o\">=</span> <span class=\"n\">key</span>\n",
        "        <span class=\"k\">for</span> <span class=\"n\">sim</span><span class=\"p\">,</span> <span class=\"n\">rest2</span><span class=\"p\">,</span> <span class=\"n\">n_common</span> <span class=\"ow\">in</span> <span class=\"nb\">sorted</span><span class=\"p\">(</span><span class=\"n\">values</span><span class=\"p\">,</span> <span class=\"n\">reverse</span><span class=\"o\">=</span><span class=\"bp\">True</span><span class=\"p\">):</span>\n",
        "            <span class=\"k\">yield</span> <span class=\"bp\">None</span><span class=\"p\">,</span> <span class=\"p\">(</span><span class=\"n\">rest1</span><span class=\"p\">,</span> <span class=\"n\">rest2</span><span class=\"p\">,</span> <span class=\"n\">sim</span><span class=\"p\">,</span> <span class=\"n\">n_common</span><span class=\"p\">)</span>\n",
        "\n",
        "<span class=\"c\">#Below MUST be there for things to work!</span>\n",
        "<span class=\"k\">if</span> <span class=\"n\">__name__</span> <span class=\"o\">==</span> <span class=\"s\">&#39;__main__&#39;</span><span class=\"p\">:</span>\n",
        "    <span class=\"n\">RestaurantSimilarities</span><span class=\"o\">.</span><span class=\"n\">run</span><span class=\"p\">()</span>\n",
        "</pre></div>\n"
       ],
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 4,
       "text": [
        "<IPython.core.display.HTML at 0x19ff5d0>"
       ]
      }
     ],
     "prompt_number": 4
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "To complete the recommender to the level of 1.7, we can now take the user's top restaurants, and repeat the process with the output of `top_similar_collector`, which could be stored in a hash table with restaurant keys and arrays of nearest neighbors."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###Submission Instructions:\n",
      "\n",
      "Restart and run your notebook one last time (you do not have to rerun the Amazon EMR script computesim.py), to make sure the output from each cell is up to date. To submit your homework, create a folder named lastname_firstinitial_hw4 and place your solutions in the folder. Double check that the file is still called HW4.ipynb, and that it contains your code. Also include the `computesim.py` script and the `output.small.local.txt` data file. Do **NOT** include the data file `output.full.emr.txt` from the larger run (its huge, so we will check your answers to 4.4 instead). Compress the folder (please use .zip compression) and submit to the CS109 dropbox in the appropriate folder. If we cannot access your work because these directions are not followed correctly, we will not grade your work!"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "###FINI\n",
      "\n",
      "You have developed all kinds of recommenders. We hope it was fun. Time constraints prevented us from going into model checking, but perhaps you would like to try that on your own. Or use S3 or a hosted database as a place to store sharded similarities. You might want to take a gander at Yelp's entire Phoenix dataset, or use the other attributes present in the data set. So many possibilities!\n",
      "\n",
      "If you'd like to learn more, please read Chris Volinksy's papers on the Netflix prize. There are also comprehensive reviews [here](http://arxiv.org/abs/1202.1112) and [here](http://www.grouplens.org/system/files/FnT%20CF%20Recsys%20Survey.pdf)."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "*css tweaks in this cell*\n",
      "<style>\n",
      "div.text_cell_render {\n",
      " line-height: 150%;\n",
      " font-size: 110%;\n",
      " width: 800px;\n",
      " margin-left:50px;\n",
      " margin-right:auto;\n",
      "}\n",
      "\n",
      "</style>"
     ]
    }
   ],
   "metadata": {}
  }
 ]
}