code/MMSB-Beer-Acctest-All.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## MMSB: Beer Survey"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "import mmsb\n",
    "import utils\n",
    "\n",
    "%matplotlib inline\n",
    "matplotlib.style.use('ggplot')\n",
    "\n",
    "from IPython.core.debugger import Tracer\n",
    "tracer = Tracer()\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('error')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>749d8459047a7c024738d643524b889a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>11</td>\n",
       "      <td>20</td>\n",
       "      <td>0</td>\n",
       "      <td>0cdd2260af00904baedc94baa0050fd6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>19</td>\n",
       "      <td>1</td>\n",
       "      <td>f933c71f15dfb64842fd9832e3734ad4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>11</td>\n",
       "      <td>1</td>\n",
       "      <td>f933c71f15dfb64842fd9832e3734ad4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>9</td>\n",
       "      <td>20</td>\n",
       "      <td>0</td>\n",
       "      <td>0cdd2260af00904baedc94baa0050fd6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    0   1  2                                 3\n",
       "0   2   5  0  749d8459047a7c024738d643524b889a\n",
       "1  11  20  0  0cdd2260af00904baedc94baa0050fd6\n",
       "2   2  19  1  f933c71f15dfb64842fd9832e3734ad4\n",
       "3   4  11  1  f933c71f15dfb64842fd9832e3734ad4\n",
       "4   9  20  0  0cdd2260af00904baedc94baa0050fd6"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = pd.read_csv('../data/all_our_ideas/2565/2565_dat.csv', header=None)\n",
    "text = pd.read_csv('../data/all_our_ideas/2565/2565_text_map.csv', header=None)[1]\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "data = data[data[3] != '1bc8052fc357986cea6bf530ff4d5d3a'] # Most prolific user"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(224, 3)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = data[[0,1,2]].values\n",
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "27"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "V = max(X[:,1]) + 1\n",
    "V"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((150, 3), (74, 3))"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.random.shuffle(X)\n",
    "X_train, X_test = X[:150], X[150:]\n",
    "X_train.shape, X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.51351351351351349"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(X_test[:,2] == 0) / float(len(X_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PolyCollection at 0x1120dfc50>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAEECAYAAAAifS8cAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIuABwFD/D3q1B4JYm2f7AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x111dbd650>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "I = pd.DataFrame(utils.get_interactions(X, V))\n",
    "plt.pcolor(I, cmap='Blues')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Tests"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0:00:01.281174\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.56756756756756754"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAEECAYAAAAifS8cAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nADAUAQ8AhiLgAcBQ/wu6w+MGMdaCbQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1124f9590>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "K = 3\n",
    "gamma, phi_pq, phi_qp, B, elbos = mmsb.train_mmsb(X_train, V, K, n_iter=400)\n",
    "ptypes = pd.DataFrame(gamma).idxmax().sort_values().index\n",
    "plt.pcolor(I.ix[ptypes][ptypes], cmap='Blues')\n",
    "probs = [gamma[:,p].dot(B).dot(gamma[:,q]) for p, q, v in X_test]\n",
    "sum(X_test[:,2] == np.round(probs)) / float(len(X_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0:00:00.962243\n",
      "0:00:01.195765\n",
      "0:00:01.247917\n",
      "0:00:01.362906\n",
      "0:00:01.460031\n",
      "0:00:01.647332\n",
      "0:00:01.872177\n",
      "0:00:01.968669\n",
      "0:00:02.186235\n",
      "0:00:02.509956\n",
      "0:00:02.693444\n",
      "0:00:03.028466\n",
      "0:00:03.314107\n",
      "0:00:03.659520\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAEECAYAAAAifS8cAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/wvA2w9XBvvcOAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112097dd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "results = []\n",
    "for K in xrange(1,15):\n",
    "    gamma, phi_pq, phi_qp, B, elbos = mmsb.train_mmsb(X_train, V, K, n_iter=400)\n",
    "    ptypes = pd.DataFrame(gamma).idxmax().sort_values().index\n",
    "    plt.pcolor(I.ix[ptypes][ptypes], cmap='Blues')\n",
    "    probs = [gamma[:,p].dot(B).dot(gamma[:,q]) for p, q, v in X_test]\n",
    "    results.append((K, sum(X_test[:,2] == np.round(probs)) / float(len(X_test))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x1122a2890>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYEAAAEECAYAAADOJIhPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112ab9290>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "x, y = zip(*results)\n",
    "plt.scatter(x=x, y=y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}