results.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "from __future__ import division\n",
    "import matplotlib.pylab as plt\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from collections import Counter"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Experiment 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Success Rates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "flist = ['log_48_r1.1','log_48_r1.2','log_100_r2','log_148_r3','log_148_r4','log_200_r5','log_200_r6','log_100_r7','log_200_r8']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "log_48_r1.1\n",
      "log_48_r1.2\n",
      "log_100_r2\n",
      "log_148_r3\n",
      "log_148_r4\n",
      "log_200_r5\n",
      "log_200_r6\n",
      "log_100_r7\n",
      "log_200_r8\n"
     ]
    }
   ],
   "source": [
    "logs = []\n",
    "for f in flist:\n",
    "    dflog = pd.read_csv('./EvalExp1/original_logs_acl2016/'+f+'_df.csv')\n",
    "    print f\n",
    "    dflog['log'] = [f]*len(dflog)\n",
    "    logs.append(dflog)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "all_logs = pd.concat(logs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{False, True, '0', 'False', 'True'}"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "set(all_logs['success'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'False', 'True'}"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "suc = []\n",
    "for s in all_logs['success']:\n",
    "    if s == True:\n",
    "        suc.append('True')\n",
    "    elif s == False:\n",
    "        suc.append('False')\n",
    "    elif s == '0':\n",
    "        suc.append('False')\n",
    "    else:\n",
    "        suc.append(s)\n",
    "        \n",
    "set(suc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'False', 'True'}"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_logs['success'] = suc\n",
    "set(all_logs['success'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gold\n",
      "293 298 329\n",
      "Overall success rate 0.98322147651\n",
      "First trial success rate 0.906040268456\n",
      "Average clicks in succseful trials 1.1228668942\n",
      "Average clicks 1.14765100671\n",
      "A 107 108 0.990740740741\n",
      "B 104 105 0.990476190476\n",
      "C 82 85 0.964705882353\n",
      "saia\n",
      "138 298 199\n",
      "Overall success rate 0.463087248322\n",
      "First trial success rate 0.322147651007\n",
      "Average clicks in succseful trials 1.44202898551\n",
      "Average clicks 2.13758389262\n",
      "A 35 98 0.357142857143\n",
      "B 60 94 0.63829787234\n",
      "C 43 106 0.405660377358\n",
      "google\n",
      "176 298 246\n",
      "Overall success rate 0.590604026846\n",
      "First trial success rate 0.41610738255\n",
      "Average clicks in succseful trials 1.39772727273\n",
      "Average clicks 2.0\n",
      "A 49 86 0.56976744186\n",
      "B 67 99 0.676767676768\n",
      "C 60 113 0.530973451327\n",
      "google_glen\n",
      "175 298 262\n",
      "Overall success rate 0.587248322148\n",
      "First trial success rate 0.375838926174\n",
      "Average clicks in succseful trials 1.49714285714\n",
      "Average clicks 2.04362416107\n",
      "A 51 102 0.5\n",
      "B 71 96 0.739583333333\n",
      "C 53 100 0.53\n"
     ]
    }
   ],
   "source": [
    "total_success = []\n",
    "first_success = []\n",
    "second_success = []\n",
    "\n",
    "a_success = []\n",
    "b_success = []\n",
    "c_success = []\n",
    "\n",
    "for system in ['gold','saia','google','google_glen']:\n",
    "    print system\n",
    "    succ = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True')])\n",
    "    first_succ = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True') & (all_logs['clicks'] == 1)])\n",
    "    second_succ = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True') & (all_logs['clicks'] <= 2)])\n",
    "\n",
    "    \n",
    "    succa = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True') \\\n",
    "                         & (all_logs['testset'] == 'A')])\n",
    "    totala = len(all_logs[(all_logs['system'] == system) & (all_logs['testset'] == 'A')])\n",
    "    succ = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True')])\n",
    "    succb = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True') \\\n",
    "                         & (all_logs['testset'] == 'B')])\n",
    "    totalb = len(all_logs[(all_logs['system'] == system) & (all_logs['testset'] == 'B')])\n",
    "    succc = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True') \\\n",
    "                         & (all_logs['testset'] == 'C')])\n",
    "    totalc = len(all_logs[(all_logs['system'] == system) & (all_logs['testset'] == 'C')])\n",
    "    \n",
    "    total = len(all_logs[(all_logs['system'] == system)])\n",
    "    clicks = sum(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True')]['clicks'])\n",
    "    allclicks = sum(all_logs[(all_logs['system'] == system) ]['clicks'])\n",
    "    \n",
    "    print succ,total,clicks\n",
    "    print \"Overall success rate\", succ/float(total)\n",
    "    print \"First trial success rate\", first_succ/float(total)\n",
    "    print \"Average clicks in succseful trials\", clicks/float(succ)\n",
    "    print \"Average clicks\",allclicks/float(total)\n",
    "    print \"A\",succa,totala,succa/float(totala)\n",
    "    print \"B\",succb,totalb,succb/float(totalb)\n",
    "    print \"C\",succc,totalc,succc/float(totalc)\n",
    "    \n",
    "    total_success.append(succ/float(total))\n",
    "    first_success.append(first_succ/float(total))\n",
    "    second_success.append(second_succ/float(total))\n",
    "    \n",
    "    a_success.append(succa/float(totala))\n",
    "    b_success.append(succb/float(totalb))\n",
    "    c_success.append(succc/float(totalc))\n",
    "    \n",
    "systems = ['gold','saia','google','google_glen']\n",
    "d = {'total_success':pd.Series(total_success,index=systems), \\\n",
    "     'first_success':pd.Series(first_success,index=systems), \\\n",
    "     'second_success':pd.Series(second_success,index=systems), \\\n",
    "     'A_success':pd.Series(a_success,index=systems),\\\n",
    "     'B_success':pd.Series(b_success,index=systems),\\\n",
    "     'C_success':pd.Series(c_success,index=systems)}\n",
    "\n",
    "evaldf = pd.DataFrame(d)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "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>A_success</th>\n",
       "      <th>B_success</th>\n",
       "      <th>C_success</th>\n",
       "      <th>first_success</th>\n",
       "      <th>second_success</th>\n",
       "      <th>total_success</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>gold</th>\n",
       "      <td>0.990741</td>\n",
       "      <td>0.990476</td>\n",
       "      <td>0.964706</td>\n",
       "      <td>0.906040</td>\n",
       "      <td>0.946309</td>\n",
       "      <td>0.983221</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>saia</th>\n",
       "      <td>0.357143</td>\n",
       "      <td>0.638298</td>\n",
       "      <td>0.405660</td>\n",
       "      <td>0.322148</td>\n",
       "      <td>0.402685</td>\n",
       "      <td>0.463087</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>google</th>\n",
       "      <td>0.569767</td>\n",
       "      <td>0.676768</td>\n",
       "      <td>0.530973</td>\n",
       "      <td>0.416107</td>\n",
       "      <td>0.533557</td>\n",
       "      <td>0.590604</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>google_glen</th>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.739583</td>\n",
       "      <td>0.530000</td>\n",
       "      <td>0.375839</td>\n",
       "      <td>0.510067</td>\n",
       "      <td>0.587248</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             A_success  B_success  C_success  first_success  second_success  \\\n",
       "gold          0.990741   0.990476   0.964706       0.906040        0.946309   \n",
       "saia          0.357143   0.638298   0.405660       0.322148        0.402685   \n",
       "google        0.569767   0.676768   0.530973       0.416107        0.533557   \n",
       "google_glen   0.500000   0.739583   0.530000       0.375839        0.510067   \n",
       "\n",
       "             total_success  \n",
       "gold              0.983221  \n",
       "saia              0.463087  \n",
       "google            0.590604  \n",
       "google_glen       0.587248  "
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "evaldf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "resultdf = pd.DataFrame(results,columns=['system','% successful trials','% 1st click success','av. clicks',\\\n",
    "                                        '% successful Test A','% successful Test B','% successful Test C'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "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>system</th>\n",
       "      <th>% successful trials</th>\n",
       "      <th>% 1st click success</th>\n",
       "      <th>av. clicks</th>\n",
       "      <th>% successful Test A</th>\n",
       "      <th>% successful Test B</th>\n",
       "      <th>% successful Test C</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>gold</td>\n",
       "      <td>0.985348</td>\n",
       "      <td>0.901099</td>\n",
       "      <td>1.153846</td>\n",
       "      <td>0.989899</td>\n",
       "      <td>0.989796</td>\n",
       "      <td>0.973684</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>saia</td>\n",
       "      <td>0.454212</td>\n",
       "      <td>0.322344</td>\n",
       "      <td>2.131868</td>\n",
       "      <td>0.373626</td>\n",
       "      <td>0.636364</td>\n",
       "      <td>0.361702</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>google</td>\n",
       "      <td>0.582418</td>\n",
       "      <td>0.413919</td>\n",
       "      <td>2.003663</td>\n",
       "      <td>0.552632</td>\n",
       "      <td>0.685393</td>\n",
       "      <td>0.518519</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>google_glen</td>\n",
       "      <td>0.589744</td>\n",
       "      <td>0.380952</td>\n",
       "      <td>2.036630</td>\n",
       "      <td>0.473684</td>\n",
       "      <td>0.755814</td>\n",
       "      <td>0.554348</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        system  % successful trials  % 1st click success  av. clicks  \\\n",
       "0         gold             0.985348             0.901099    1.153846   \n",
       "1         saia             0.454212             0.322344    2.131868   \n",
       "2       google             0.582418             0.413919    2.003663   \n",
       "3  google_glen             0.589744             0.380952    2.036630   \n",
       "\n",
       "   % successful Test A  % successful Test B  % successful Test C  \n",
       "0             0.989899             0.989796             0.973684  \n",
       "1             0.373626             0.636364             0.361702  \n",
       "2             0.552632             0.685393             0.518519  \n",
       "3             0.473684             0.755814             0.554348  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resultdf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Success rate over time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "s_time = [0]*200\n",
    "\n",
    "for f in ['log_200_r5','log_200_r6','log_200_r8']:\n",
    "    dflog = pd.read_csv('./EvalExp1/original_logs_acl2016/'+f+'_df.csv')\n",
    "    for s in range(200):\n",
    "        val = dflog['success'].iloc[s]\n",
    "        if val == 'True':\n",
    "            s_time[s] += 1\n",
    "        elif val == 'False':\n",
    "            pass\n",
    "        else:\n",
    "            s_time[s] += int(val)        \n",
    "\n",
    "y_s_time = []\n",
    "for s in range(200):\n",
    "    y_s_time.append(sum(s_time[:s+1])/(3*(s+1)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x114262050>]"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAEACAYAAAC08h1NAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n8/8BTE6hGQrV6CMAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11419ffd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(1,201),y_s_time)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Experiment 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* first read in all the log files\n",
    "\n",
    "* you might need to adapt the file names/path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "log_ia_300_r1\n",
      "log_ia_300_r2\n"
     ]
    }
   ],
   "source": [
    "flist = ['log_ia_300_r1','log_ia_300_r2']\n",
    "logs = []\n",
    "for f in flist:\n",
    "    dflog = pd.read_csv('./EvalExp2/original_logs_acl2016/'+f+'_df.csv')\n",
    "    print f\n",
    "    dflog['log'] = [f]*len(dflog)\n",
    "    dflog['system'] = ['context-aware']*len(dflog)\n",
    "    logs.append(dflog)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "log_ia_bl_200_r3\n",
      "log_ia_bl_200_r4\n",
      "log_ia_bl_200_r5\n"
     ]
    }
   ],
   "source": [
    "flist = ['log_ia_bl_200_r3','log_ia_bl_200_r4','log_ia_bl_200_r5']\n",
    "for f in flist:\n",
    "    dflog = pd.read_csv('./EvalExp2/original_logs_acl2016/'+f+'_df.csv')\n",
    "    print f\n",
    "    dflog['log'] = [f]*len(dflog)\n",
    "    dflog['system'] = ['pattern-based']*len(dflog)\n",
    "    logs.append(dflog)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "all_logs = pd.concat(logs)\n",
    "suc = []\n",
    "for s in all_logs['success']:\n",
    "    if s == True:\n",
    "        suc.append('True')\n",
    "    elif s == False:\n",
    "        suc.append('False')\n",
    "    elif s == '0':\n",
    "        suc.append('False')\n",
    "    else:\n",
    "        suc.append(s)\n",
    "all_logs['success'] = suc\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "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>Unnamed: 0</th>\n",
       "      <th>clicks</th>\n",
       "      <th>file</th>\n",
       "      <th>lastClick</th>\n",
       "      <th>log</th>\n",
       "      <th>ref1</th>\n",
       "      <th>ref2</th>\n",
       "      <th>ref3</th>\n",
       "      <th>refexp</th>\n",
       "      <th>reftype</th>\n",
       "      <th>region</th>\n",
       "      <th>start</th>\n",
       "      <th>success</th>\n",
       "      <th>system</th>\n",
       "      <th>testset</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>8756</td>\n",
       "      <td>2.007509</td>\n",
       "      <td>log_ia_300_r1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>('guy in the middle in front', 'guy in the mid...</td>\n",
       "      <td>test</td>\n",
       "      <td>10</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>context-aware</td>\n",
       "      <td>T</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>9415</td>\n",
       "      <td>2.420646</td>\n",
       "      <td>log_ia_300_r1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>('orange thing on left', 'doorway or house thi...</td>\n",
       "      <td>a2</td>\n",
       "      <td>12</td>\n",
       "      <td>2.186790</td>\n",
       "      <td>False</td>\n",
       "      <td>context-aware</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>7063</td>\n",
       "      <td>2.582398</td>\n",
       "      <td>log_ia_300_r1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>('bottom left', 'ground bottom left corner', '...</td>\n",
       "      <td>a1</td>\n",
       "      <td>4</td>\n",
       "      <td>2.473041</td>\n",
       "      <td>True</td>\n",
       "      <td>context-aware</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>37098</td>\n",
       "      <td>2.819412</td>\n",
       "      <td>log_ia_300_r1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>('car in middle left', 'van or car in middle l...</td>\n",
       "      <td>a2</td>\n",
       "      <td>3</td>\n",
       "      <td>2.626567</td>\n",
       "      <td>True</td>\n",
       "      <td>context-aware</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>40188</td>\n",
       "      <td>2.953601</td>\n",
       "      <td>log_ia_300_r1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>('in front', 'guy in front center', 'animal or...</td>\n",
       "      <td>a1</td>\n",
       "      <td>1</td>\n",
       "      <td>2.873322</td>\n",
       "      <td>True</td>\n",
       "      <td>context-aware</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0  clicks   file  lastClick            log ref1 ref2 ref3  \\\n",
       "0           0       1   8756   2.007509  log_ia_300_r1  NaN  NaN  NaN   \n",
       "1           1       3   9415   2.420646  log_ia_300_r1  NaN  NaN  NaN   \n",
       "2           2       2   7063   2.582398  log_ia_300_r1  NaN  NaN  NaN   \n",
       "3           3       2  37098   2.819412  log_ia_300_r1  NaN  NaN  NaN   \n",
       "4           4       1  40188   2.953601  log_ia_300_r1  NaN  NaN  NaN   \n",
       "\n",
       "                                              refexp reftype  region  \\\n",
       "0  ('guy in the middle in front', 'guy in the mid...    test      10   \n",
       "1  ('orange thing on left', 'doorway or house thi...      a2      12   \n",
       "2  ('bottom left', 'ground bottom left corner', '...      a1       4   \n",
       "3  ('car in middle left', 'van or car in middle l...      a2       3   \n",
       "4  ('in front', 'guy in front center', 'animal or...      a1       1   \n",
       "\n",
       "      start success         system testset  \n",
       "0  0.000000    True  context-aware       T  \n",
       "1  2.186790   False  context-aware       A  \n",
       "2  2.473041    True  context-aware       A  \n",
       "3  2.626567    True  context-aware       A  \n",
       "4  2.873322    True  context-aware       A  "
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_logs.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* calculate various measures of referential success"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "pattern-based\n",
      "488 603 766\n",
      "Overall success rate 0.809286898839\n",
      "First trial success rate 0.467661691542\n",
      "Average clicks in succseful trials 1.56967213115\n",
      "Average clicks 1.83250414594\n",
      "A 160 198 0.808080808081\n",
      "B 167 198 0.843434343434\n",
      "C 158 204 0.774509803922\n",
      "context-aware\n",
      "518 602 799\n",
      "Overall success rate 0.860465116279\n",
      "First trial success rate 0.504983388704\n",
      "Average clicks in succseful trials 1.54247104247\n",
      "Average clicks 1.74086378738\n",
      "A 172 200 0.86\n",
      "B 175 200 0.875\n",
      "C 169 200 0.845\n"
     ]
    }
   ],
   "source": [
    "total_success = []\n",
    "first_success = []\n",
    "second_success = []\n",
    "\n",
    "a_success = []\n",
    "b_success = []\n",
    "c_success = []\n",
    "\n",
    "for system in ['pattern-based','context-aware']:\n",
    "    print system\n",
    "    succ = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True')])\n",
    "    first_succ = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True') & (all_logs['clicks'] == 1)])\n",
    "    second_succ = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True') & (all_logs['clicks'] <= 2)])\n",
    "\n",
    "    \n",
    "    succa = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True') \\\n",
    "                         & (all_logs['testset'] == 'A')])\n",
    "    totala = len(all_logs[(all_logs['system'] == system) & (all_logs['testset'] == 'A')])\n",
    "    succ = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True')])\n",
    "    succb = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True') \\\n",
    "                         & (all_logs['testset'] == 'B')])\n",
    "    totalb = len(all_logs[(all_logs['system'] == system) & (all_logs['testset'] == 'B')])\n",
    "    succc = len(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True') \\\n",
    "                         & (all_logs['testset'] == 'C')])\n",
    "    totalc = len(all_logs[(all_logs['system'] == system) & (all_logs['testset'] == 'C')])\n",
    "    \n",
    "    total = len(all_logs[(all_logs['system'] == system)])\n",
    "    clicks = sum(all_logs[(all_logs['system'] == system) & (all_logs['success'] == 'True')]['clicks'])\n",
    "    allclicks = sum(all_logs[(all_logs['system'] == system) ]['clicks'])\n",
    "    \n",
    "    print succ,total,clicks\n",
    "    print \"Overall success rate\", succ/float(total)\n",
    "    print \"First trial success rate\", first_succ/float(total)\n",
    "    print \"Average clicks in succseful trials\", clicks/float(succ)\n",
    "    print \"Average clicks\",allclicks/float(total)\n",
    "    print \"A\",succa,totala,succa/float(totala)\n",
    "    print \"B\",succb,totalb,succb/float(totalb)\n",
    "    print \"C\",succc,totalc,succc/float(totalc)\n",
    "    \n",
    "    total_success.append(succ/float(total))\n",
    "    first_success.append(first_succ/float(total))\n",
    "    second_success.append(second_succ/float(total))\n",
    "    \n",
    "    a_success.append(succa/float(totala))\n",
    "    b_success.append(succb/float(totalb))\n",
    "    c_success.append(succc/float(totalc))\n",
    "    \n",
    "systems = ['pattern-based','context-aware']\n",
    "d = {'total_success':pd.Series(total_success,index=systems), \\\n",
    "     'first_success':pd.Series(first_success,index=systems), \\\n",
    "     'second_success':pd.Series(second_success,index=systems), \\\n",
    "     'A_success':pd.Series(a_success,index=systems),\\\n",
    "     'B_success':pd.Series(b_success,index=systems),\\\n",
    "     'C_success':pd.Series(c_success,index=systems)}\n",
    "\n",
    "evaldf = pd.DataFrame(d)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "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>A_success</th>\n",
       "      <th>B_success</th>\n",
       "      <th>C_success</th>\n",
       "      <th>first_success</th>\n",
       "      <th>second_success</th>\n",
       "      <th>total_success</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>pattern-based</th>\n",
       "      <td>0.808081</td>\n",
       "      <td>0.843434</td>\n",
       "      <td>0.77451</td>\n",
       "      <td>0.467662</td>\n",
       "      <td>0.691542</td>\n",
       "      <td>0.809287</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>context-aware</th>\n",
       "      <td>0.860000</td>\n",
       "      <td>0.875000</td>\n",
       "      <td>0.84500</td>\n",
       "      <td>0.504983</td>\n",
       "      <td>0.749169</td>\n",
       "      <td>0.860465</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               A_success  B_success  C_success  first_success  second_success  \\\n",
       "pattern-based   0.808081   0.843434    0.77451       0.467662        0.691542   \n",
       "context-aware   0.860000   0.875000    0.84500       0.504983        0.749169   \n",
       "\n",
       "               total_success  \n",
       "pattern-based       0.809287  \n",
       "context-aware       0.860465  "
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "evaldf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Learning Curves"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* now calculate success rates over time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "s_time = [0]*200\n",
    "\n",
    "for f in ['log_ia_300_r1','log_ia_300_r2']:\n",
    "    dflog = pd.read_csv('./EvalExp2/original_logs_Acl2016/'+f+'_df.csv')\n",
    "    for s in range(200):\n",
    "        val = dflog['success'].iloc[s]\n",
    "        if val == 'True':\n",
    "            s_time[s] += 1\n",
    "        elif val == 'False':\n",
    "            pass\n",
    "        else:\n",
    "            s_time[s] += int(val)        \n",
    "\n",
    "y_s_time = []\n",
    "for s in range(200):\n",
    "    y_s_time.append(sum(s_time[:s+1])/(2*(s+1)))\n",
    "    \n",
    "s_time = [0]*200\n",
    "\n",
    "for f in ['log_ia_bl_200_r3','log_ia_bl_200_r4','log_ia_bl_200_r5']:\n",
    "    dflog = pd.read_csv('./EvalExp2/original_logs_acl2016/'+f+'_df.csv')\n",
    "    for s in range(200):\n",
    "        val = dflog['success'].iloc[s]\n",
    "        if val == 'True':\n",
    "            s_time[s] += 1\n",
    "        elif val == 'False':\n",
    "            pass\n",
    "        else:\n",
    "            s_time[s] += int(val)        \n",
    "\n",
    "y_s_time_bl = []\n",
    "for s in range(200):\n",
    "    y_s_time_bl.append(sum(s_time[:s+1])/(3*(s+1)))\n",
    "    \n",
    "s_time = [0]*200\n",
    "\n",
    "for f in ['log_200_r5','log_200_r6','log_200_r8']:\n",
    "    dflog = pd.read_csv('./EvalExp1/original_logs_acl2016/'+f+'_df.csv')\n",
    "    for s in range(200):\n",
    "        val = dflog['success'].iloc[s]\n",
    "        if val == 'True':\n",
    "            s_time[s] += 1\n",
    "        elif val == 'False':\n",
    "            pass\n",
    "        else:\n",
    "            s_time[s] += int(val)        \n",
    "\n",
    "y_s_time_no = []\n",
    "for s in range(200):\n",
    "    y_s_time_no.append(sum(s_time[:s+1])/(3*(s+1)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEUCAYAAAA4DAFkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nKT31IeQZab0AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1144ca150>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(1,201),y_s_time,label=\"Context installments\")\n",
    "plt.plot(range(1,201),y_s_time_bl,linestyle=\"dashed\",lw=2,label=\"Simple installments\")\n",
    "plt.plot(range(1,201),y_s_time_no,linestyle=\"dashdot\",lw=3,label=\"One-shot\")\n",
    "plt.xlabel(\"# Game\",fontsize=15)\n",
    "plt.ylabel(\"% Accuracy\",fontsize=15)\n",
    "plt.ylim(0.5,1.2)\n",
    "plt.legend(fontsize=15)\n",
    "plt.savefig(\"learningcurves.pdf\")"
   ]
  },
  {
   "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": "2.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}