Programación/Notebooks/jupyter_sessio_4.ipynb

{
 "metadata": {
  "name": "",
  "signature": "sha256:cd693108f34ecadc38614060555333e64a88ad4ae85d05d3ee1a36da6f54b2a6"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Jupyter Notebook Sessi\u00f3 4"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "En aquesta Jupyter Notebok de la sessi\u00f3 4, mostrarem quins son els resultats dels nostres dos algoritmes obtinguts fins ara, tant el classificador com el ranker han estat implementats de forma aleat\u00f2ria es a dir sense tenir en compte les caracter\u00edstiques de les imatges."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import os\n",
      "os.chdir('C:\\Users\\Universitat\\Dropbox\\ProjecteGDSA\\Repositori')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 1
    },
    {
     "cell_type": "heading",
     "level": 4,
     "metadata": {},
     "source": [
      "Importerm les funcions necessaries"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from get_params import get_params\n",
      "from eval_random_rank import eval_rankings\n",
      "from eval_random_class import eval_classification\n",
      "from eval_random_class import plot_confusion_matrix\n",
      "import cv2\n",
      "import numpy as np\n",
      "import matplotlib.pyplot as plt\n",
      "import matplotlib.image as mpimg\n",
      "import warnings\n",
      "warnings.filterwarnings('ignore')\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 2
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "1-Classificador aleat\u00f2ri"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Comen\u00e7arem per mostrar els resultats d'una classificaci\u00f3 aleat\u00f2ria, on per cada imatge de validaci\u00f3 el classificador predir\u00e0 una de les 13 possibles classes aleatoriament. Per poder mostrar els resultats d'una forma visual utilitzarem una matriu de confusi\u00f3, en la que si obtinguessim una classificaci\u00f3 perfecta tots els valors alts estarien situats a la diagonal."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "params=get_params()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 3
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "f1, precision, recall, accuracy,cm, labels = eval_classification(params)\n",
      "plot_confusion_matrix(cm, labels,normalize = True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAUoAAAEnCAYAAADPQwqBAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0xd52ff60>"
       ]
      }
     ],
     "prompt_number": 4
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Tamb\u00e9 podem observar altres mesures com:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print \"F1 per class:\"\n",
      "print zip(labels,f1)\n",
      "print \"Precision per class:\"\n",
      "print zip(labels,precision)\n",
      "print \"Recall per class:\"\n",
      "print zip(labels,recall)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "F1 per class:\n",
        "[('ajuntament', 0.080000000000000002), ('castell_cartoixa', 0.0), ('catedral', 0.13793103448275862), ('desconegut', 0.25862068965517243), ('dona_treballadora', 0.0), ('escola_enginyeria', 0.090909090909090898), ('estacio_nord', 0.086956521739130432), ('farmacia_albinyana', 0.095238095238095247), ('masia_freixa', 0.0), ('mercat_independencia', 0.0), ('mnactec', 0.11111111111111112), ('societat_general', 0.11111111111111112), ('teatre_principal', 0.0)]\n",
        "Precision per class:\n",
        "[('ajuntament', 0.066666666666666666), ('castell_cartoixa', 0.0), ('catedral', 0.10526315789473684), ('desconegut', 0.26785714285714285), ('dona_treballadora', 0.0), ('escola_enginyeria', 0.083333333333333329), ('estacio_nord', 0.076923076923076927), ('farmacia_albinyana', 0.090909090909090912), ('masia_freixa', 0.0), ('mercat_independencia', 0.0), ('mnactec', 0.125), ('societat_general', 0.125), ('teatre_principal', 0.0)]\n",
        "Recall per class:\n",
        "[('ajuntament', 0.10000000000000001), ('castell_cartoixa', 0.0), ('catedral', 0.20000000000000001), ('desconegut', 0.25), ('dona_treballadora', 0.0), ('escola_enginyeria', 0.10000000000000001), ('estacio_nord', 0.10000000000000001), ('farmacia_albinyana', 0.10000000000000001), ('masia_freixa', 0.0), ('mercat_independencia', 0.0), ('mnactec', 0.10000000000000001), ('societat_general', 0.10000000000000001), ('teatre_principal', 0.0)]\n"
       ]
      }
     ],
     "prompt_number": 5
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print \"Average F1:\", np.mean(f1)\n",
      "print \"Average Precision:\", np.mean(precision)\n",
      "print \"Average Recall:\", np.mean(recall)\n",
      "print \"Accuracy:\", accuracy"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Average F1: 0.0747598195574\n",
        "Average Precision: 0.0723809591218\n",
        "Average Recall: 0.0807692307692\n",
        "Accuracy: 0.127777777778\n"
       ]
      }
     ],
     "prompt_number": 6
    },
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "2-Ranker aleat\u00f2ri"
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "A continuaci\u00f3 analitzarem l'avaluaci\u00f3 del ranker aleatori, per fer-ho no tindrem en compte la classe \"desconegut\" ja que no ens interessa. Primer de tot veiem la precisi\u00f3 en mitjana per cada una de les 120 imatges de validaci\u00f3, despr\u00e9s fem la mitjana de les precisions de les 120 imatges."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "ap_list=eval_rankings(params)\n",
      "print \"-Llista de Average Precission: \"\n",
      "print ap_list\n",
      "print \"\\n\"\n",
      "print \"-Mean Average Precission: \"\n",
      "print np.mean(ap_list)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "-Llista de Average Precission: \n",
        "[0.07859852039046294, 0.06537121467682715, 0.07165622093848534, 0.09687447749546159, 0.08207881335164399, 0.06728840201024909, 0.06716404967957998, 0.06698544243870602, 0.06965448998486887, 0.05623328859033241, 0.060735756412755255, 0.05005221515782393, 0.061034604416855945, 0.05814681293097629, 0.05243889962319601, 0.07171546988413323, 0.046946930174976795, 0.0362795850046182, 0.10183844194023889, 0.05993562803679404, 0.0315527362661492, 0.05684075661903979, 0.09482908574388609, 0.04781105878196228, 0.06044412517364156, 0.05396029838850422, 0.08022122535746494, 0.07219364324788898, 0.08967689451762575, 0.06426885116682629, 0.04540326479661251, 0.05820372181924293, 0.06065104273559886, 0.03569374380508811, 0.09994415936348447, 0.0884036025256929, 0.09160291589511511, 0.0732697920807996, 0.02669216611278252, 0.06525424133356397, 0.09119486374939653, 0.05105426152195473, 0.05952281054037062, 0.05187363827709223, 0.048344051890933784, 0.04123888762279534, 0.0829880110560984, 0.10906049986857984, 0.06031817369968397, 0.060724148191156964, 0.05125168643236854, 0.03947321978504457, 0.06108305493179208, 0.04040439449562759, 0.10955720192647525, 0.0580518211313503, 0.08346540632563502, 0.06787740042969864, 0.09995896048779482, 0.11468382503715006, 0.06623106627584904, 0.052207664296069024, 0.10483156488334994, 0.08949563295679781, 0.05496886647838983, 0.08146458354099835, 0.0631879554646586, 0.05676055576058808, 0.061149543033386336, 0.05880618555256469, 0.06097316961847281, 0.049957863614231594, 0.09813909196697936, 0.06913052112783312, 0.06616923287851893, 0.04520752184142624, 0.07886561612230955, 0.06765007240073358, 0.07430452118204699, 0.06432052726866434, 0.060399222105888696, 0.06362876037119945, 0.0734632656664682, 0.0620662226066712, 0.08074548404998105, 0.07939985084456114, 0.07415755766453479, 0.09735229678143377, 0.05379574270231667, 0.10212586720950964, 0.08292708023832455, 0.09184593664037487, 0.04836095622282559, 0.07925688276363496, 0.07775502348626989, 0.07714535501615817, 0.06443718470887091, 0.0741328544187178, 0.0412812235390733, 0.08907544396638284, 0.054197087874227765, 0.15284518654109205, 0.048687942427088085, 0.0788423814453046, 0.0475095600618452, 0.05572388579848288, 0.045687650985037236, 0.04338202910401873, 0.05241601731270019, 0.09291716088409842, 0.06902656660448069, 0.07488679445496715, 0.04906770265644107, 0.06607416870980458, 0.06387185993387014, 0.06397523937382994, 0.05214451812917159, 0.04953737744218902, 0.08378007684733084, 0.05888607295501391]\n",
        "\n",
        "\n",
        "-Mean Average Precission: \n",
        "0.0678392179423\n"
       ]
      }
     ],
     "prompt_number": 7
    }
   ],
   "metadata": {}
  }
 ]
}