predict.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Warning: Maximum number of iterations has been exceeded.\n",
      "Final Cost= [ 0.24021978]\n",
      "accuracy= 97.2972972973\n"
     ]
    }
   ],
   "source": [
    "import csv\n",
    "import numpy as np\n",
    "import scipy.optimize, scipy.special\n",
    "from scipy.optimize import fmin_bfgs\n",
    "from matplotlib import pyplot, cm\n",
    "from mpl_toolkits.mplot3d import Axes3D\n",
    "\n",
    "with open('../USOpen-men-2013.csv','rb') as csvfile:\n",
    "    spamreader = csv.reader(csvfile)\n",
    "    arr=[]\n",
    "    cnt=0\n",
    "    for row in spamreader:\n",
    "         if cnt==0:\n",
    "            cnt=cnt+1  \n",
    "         else:\n",
    "            arr.append(row)\n",
    "    x=[]\n",
    "    y=[]\n",
    "#     x contains the features\n",
    "#     y contains the class\n",
    "    for row in arr:\n",
    "        y.append([np.float64(row[3])])\n",
    "    for row in arr:\n",
    "        cnt=0\n",
    "        brr=[]\n",
    "        for i in row:\n",
    "            if cnt<6:\n",
    "                cnt=cnt+1\n",
    "            else:\n",
    "                brr.append(i)\n",
    "        x.append(brr) \n",
    "m=len(x) #no of training examples\n",
    "n=len(x[0]) # no of features\n",
    "col=np.ones((m,1))\n",
    "X=np.hstack((col,x))\n",
    "theta=np.zeros((n+1,1))\n",
    "# print len(theta)\n",
    "for i in range(m):\n",
    "    for j in range(n+1):\n",
    "        if X[i][j]=='' or X[i][j]=='NA':\n",
    "            X[i][j]='0.0'\n",
    "X=np.array(X,dtype=np.float64) \n",
    "y=np.array(y,dtype=np.float64)\n",
    "theta=np.array(theta,dtype=np.float64)\n",
    "ss=compute_cost(theta,X,y)\n",
    "result=scipy.optimize.fmin( compute_cost, x0=theta, args=(X, y), maxiter=1500, full_output=True )\n",
    "# print result[0] # display the parameters theta\n",
    "for i in range(n+1):\n",
    "    theta[i]=[result[0][i]]\n",
    "hyp=sigmoid(X.dot(theta))\n",
    "print 'Final Cost=',compute_cost(theta,X,y) # final reduced cost \n",
    "cnt=0\n",
    "for i in range(n+1):\n",
    "    if hyp[i]>0.5:\n",
    "        if y[i]==1:\n",
    "            cnt=cnt+1\n",
    "    else:\n",
    "        if y[i]==0:\n",
    "            cnt=cnt+1\n",
    "print 'accuracy=',100.0*(float(cnt)/float(n+1))            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def sigmoid(z):\n",
    "    return 1.0/(1+np.exp(-z))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def compute_cost(theta,X,y): #computes cost given predicted and actual values\n",
    "    cnt=0.0\n",
    "    grad=np.zeros((n+1,1))\n",
    "    hyp=sigmoid(X.dot(theta))\n",
    "    z= hyp - y\n",
    "    J=0\n",
    "    for i in range(n+1):\n",
    "        for j in range(m):\n",
    "            grad[i]=grad[i]+((hyp[i]-y[i]))*X[j][i]\n",
    "        grad[i]=grad[i]/m    \n",
    "    for i in range(m):\n",
    "        J=J+(y[i]*np.log(hyp[i]))+((1-y[i])*np.log(1-hyp[i]))\n",
    "    J=-J/m\n",
    "    return J"
   ]
  }
 ],
 "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": "2.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}