svm-regression-poly-kernel.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variance score of poly: -5013749.615741554\n",
      "Mean Squared Error of test data for poly:  40020871.5817\n",
      "Mean Squared Error of train data for poly:  8535050.75802\n",
      "Real new dose :  [  6.   8.   6.   5.  14.   5.   6.  10.   4.   8.   6.   5.   8.  12.   4.]\n",
      "Predicted new dose using poly:  [   740.79418619  -1420.29038755  -4470.61439888    197.49498184\n",
      "  16470.1240495    9432.23787363    734.14746983  -4798.34326689\n",
      "  -5502.52993481  -3281.85732786  -1933.43208735    310.58270224\n",
      "  -6603.45230619 -10169.33995287   1403.29090317]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAcEAAAEeCAYAAADy064SAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5854383860>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt \n",
    "import numpy as np \n",
    "from sklearn.svm import SVR\n",
    "from sklearn import datasets, metrics\n",
    "import csv\n",
    "from copy import deepcopy\n",
    "\n",
    "def load_csv(filename):\n",
    "    raw_data = open(filename, 'rt')\n",
    "    reader = csv.reader(raw_data, delimiter=',', quoting=csv.QUOTE_NONE)\n",
    "    x = list(reader)\n",
    "    data = np.array(x).astype('float')\n",
    "    return data\n",
    "    \n",
    "def splitDataset(dataset, splitratio):\n",
    "    trainsize = int(np.round(dataset.shape[0]*splitratio))\n",
    "    trainset = np.zeros((trainsize,dataset.shape[1]))#array to store the training set.\n",
    "    testset = deepcopy(dataset)#create a copy of the dataset in test set.\n",
    "    for numsamples in range(trainsize):\n",
    "        indx = np.random.randint(0,testset.shape[0])#random index generation\n",
    "        trainset[numsamples,:] = testset[indx,:]#adding the randomly selected data vector to the training set\n",
    "        testset = np.delete(testset, indx, axis = 0)#delete the selected observation from the test set.\n",
    "    return trainset,testset\n",
    "\n",
    "data = load_csv('preprocessed_data.csv')\n",
    "trainset, testset = splitDataset(data,0.7)\n",
    "\n",
    "x_train = trainset[:,:-1]\n",
    "y_train = trainset[:,-1]\n",
    "\n",
    "x_test = testset[:,:-1]\n",
    "y_test = testset[:,-1]\n",
    "  \n",
    "# create linear regression object \n",
    "reg = SVR(kernel='poly', C=1e3, gamma=0.1)\n",
    "  \n",
    "# train the model using the training sets \n",
    "reg.fit(x_train, y_train)\n",
    "\n",
    "# variance score: 1 means perfect prediction \n",
    "print('Variance score of poly: {}'.format(reg.score(x_test, y_test)))\n",
    "\n",
    "# mean squared error\n",
    "print('Mean Squared Error of test data for poly: ', metrics.mean_squared_error(y_test, reg.predict(x_test)))\n",
    "print('Mean Squared Error of train data for poly: ', metrics.mean_squared_error(y_train, reg.predict(x_train)))\n",
    "\n",
    "# real and predicted new dose\n",
    "print('Real new dose : ', y_test)\n",
    "print('Predicted new dose using poly: ', reg.predict(x_test))\n",
    "\n",
    "# plot for residual error of poly\n",
    "  \n",
    "## setting plot style \n",
    "plt.style.use('fivethirtyeight') \n",
    "  \n",
    "## plotting residual errors in training data \n",
    "plt.scatter(reg.predict(x_train), reg.predict(x_train) - y_train, \n",
    "            color = \"green\", s = 10, label = 'Train data') \n",
    "  \n",
    "## plotting residual errors in test data \n",
    "plt.scatter(reg.predict(x_test), reg.predict(x_test) - y_test, \n",
    "            color = \"blue\", s = 10, label = 'Test data') \n",
    "  \n",
    "## plotting line for zero residual error \n",
    "plt.hlines(y = 0, xmin = 0, xmax = 50, linewidth = 2) \n",
    "  \n",
    "## plotting legend \n",
    "plt.legend(loc = 'upper right') \n",
    "  \n",
    "## plot title \n",
    "plt.title(\"Residual errors\") \n",
    "  \n",
    "## function to show plot \n",
    "plt.show()\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}