baseline.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.cross_validation import cross_val_score\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x10ad75d50>"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD7CAYAAABOrvnfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x107456e50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD7CAYAAABOrvnfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkP8Hrlc8fZYqyd0AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10a52ca90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_data = np.load('train_data.npz')\n",
    "validation_data = np.load('validation_data.npz')\n",
    "\n",
    "plt.matshow(train_data['X'][2224].reshape(28, 28), cmap=plt.get_cmap('Greys'))\n",
    "plt.matshow(train_data['X'][2225].reshape(28, 28), cmap=plt.get_cmap('Greys'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def test_classifier(cl, train_x, train_y):\n",
    "    print 'Train accuracy: ', cl.score(train_x, train_y)\n",
    "    print 'Cross-validation accuracy: ', cross_val_score(cl, train_x, train_y, cv=10).mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train accuracy:  0.860515021459\n",
      "Cross-validation accuracy:  0.824034334764\n"
     ]
    }
   ],
   "source": [
    "cl = KNeighborsClassifier()\n",
    "cl.fit(train_data['X'], train_data['y'])\n",
    "test_classifier(cl, train_data['X'], train_data['y'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "train_x = train_data['X'][1950:].copy()\n",
    "train_y = train_data['y'][1950:].copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train accuracy:  0.505263157895\n",
      "Cross-validation accuracy:  0.305263157895\n"
     ]
    }
   ],
   "source": [
    "cl = KNeighborsClassifier()\n",
    "cl.fit(train_x, train_y)\n",
    "test_classifier(cl, train_x, train_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x10f148490>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD7CAYAAABOrvnfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10623c7d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD7CAYAAABOrvnfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10b6435d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def binarize(x):\n",
    "    x[x > 0] = 255\n",
    "    return x\n",
    "\n",
    "train_x = binarize(train_x)\n",
    "validation_x = binarize(validation_data['X']) # not forgetting about preprocessing validation data\n",
    "\n",
    "plt.matshow(train_x[2224 - 1950].reshape(28, 28), cmap=plt.get_cmap('Greys'))\n",
    "plt.matshow(train_x[2225 - 1950].reshape(28, 28), cmap=plt.get_cmap('Greys'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train accuracy:  0.823684210526\n",
      "Cross-validation accuracy:  0.726315789474\n"
     ]
    }
   ],
   "source": [
    "cl = KNeighborsClassifier()\n",
    "cl.fit(train_x, train_y)\n",
    "test_classifier(cl, train_x, train_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "fout = open('answer.csv', 'w')\n",
    "fout.write('id,answer\\n')\n",
    "\n",
    "i = 0\n",
    "for y in cl.predict(validation_x):\n",
    "    fout.write('%d,%d\\n' % (i, y))\n",
    "    i += 1\n",
    "\n",
    "fout.close()"
   ]
  },
  {
   "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": "2.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}