Parson_Window_Classifier.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import math\n",
    "from mpl_toolkits.mplot3d import Axes3D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Given Data\n",
    "x1_w1 = np.array([0.28,0.07,1.54,-0.44,-0.81,1.52,2.2,0.91,0.65,-0.26])\n",
    "x2_w1 = np.array([1.31,0.58,2.01,1.18,0.21,3.16,2.42,1.94,1.93,0.82])\n",
    "x3_w1 = np.array([-6.2,-0.78,-1.63,-4.32,5.73,2.77,-0.19,6.21,4.38,-0.96])\n",
    "fig = plt.figure()\n",
    "ax = Axes3D(fig)\n",
    "ax.scatter(x1_w1,x2_w1,x3_w1,s=50,alpha=0.9);\n",
    "x1_w2 = np.array([0.011,1.27,0.13,-0.21,-2.18,0.34,-1.38,-0.12,-1.44,0.26])\n",
    "x2_w2 = np.array([1.03,1.28,3.12,1.23,1.39,1.96,0.94,0.82,2.31,1.94])\n",
    "x3_w2 = np.array([-0.21,0.08,0.16,-0.11,-0.19,-0.16,0.45,0.17,0.14,0.08])\n",
    "ax.scatter(x1_w2,x2_w2,x3_w2,s=50,alpha=0.9);\n",
    "x1_w3 = np.array([1.36,1.41,1.22,2.46,0.68,2.51,0.6,0.64,0.85,0.66])\n",
    "x2_w3 = np.array([2.17,1.45,0.99,2.19,0.79,3.22,2.44,0.13,0.58,0.51])\n",
    "x3_w3 = np.array([0.14,-0.38,0.69,1.31,0.87,1.35,0.92,0.97,0.99,0.88])\n",
    "ax.scatter(x1_w3,x2_w3,x3_w3,s=50,alpha=0.9);\n",
    "X_W1 = np.stack((x1_w1,x2_w1,x3_w1),axis=0)\n",
    "X_W2 = np.stack((x1_w2,x2_w2,x3_w2),axis=0)\n",
    "X_W3 = np.stack((x1_w3,x2_w3,x3_w3),axis=0)\n",
    "X_W = np.concatenate((X_W1,X_W2,X_W3),axis=1) #all training examples"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Function for calculating P(given point) using parzon window:Parzon Window estimate"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**ϕ((x − xi)/h) ∝ Exp[−(x − xi)^t (x − xi)/(2h^2)]**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "def Parzon(point,X,h):\n",
    "    k = 0\n",
    "    for i in range(len(X)):\n",
    "        k = k + np.exp(-np.dot((point-X[:,i]).T,(point-X[:,i]))/(2*(h**2)))\n",
    "    px_w = k/(len(X)*(h**3))\n",
    "    return px_w"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### a) Write a program to classify an arbitrary test point x based on the Parzen window estimates.\n",
    "**Train your classifier using the three-dimensional data from three categories in the table above.\n",
    "Set h = 1 and classify the following three points: (0.50, 1.0, 0.0)t, (0.31, 1.51, −0.50)t and (−0.3, 0.44, −0.1)t.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "h = 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Define function to find class of given point**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "def Parson_classifier(point,X_W1,X_W2,X_W3,X_W,h):\n",
    "    p_of_point_given_class_w1 = Parzon(point,X_W1,h)\n",
    "     #p(point1|w1)\n",
    "\n",
    "    p_of_point_given_class_w2 = Parzon(point,X_W2,h)\n",
    "     #p(point1|w2)\n",
    "\n",
    "    p_of_point_given_class_w3 = Parzon(point,X_W3,h)\n",
    "     #p(point1|w3)\n",
    "\n",
    "    total_p_point = Parzon(point,X_W,h) # X_W =all training examples\n",
    "    #p(point1)\n",
    "\n",
    "    p_of_class_w1_given_point = p_of_point_given_class_w1/total_p_point\n",
    "    ##p(w1|point1) = p(point1|w1)/p(point1)\n",
    "    print(\"p(class_w1|given_point):\",p_of_class_w1_given_point)\n",
    "    p_of_class_w2_given_point = p_of_point_given_class_w2/total_p_point\n",
    "    ##p(w2|point1)\n",
    "    print(\"p(class_w2|given_point):\",p_of_class_w2_given_point)\n",
    "    p_of_class_w3_given_point = p_of_point1_given_class_w3/total_p_point\n",
    "    ##p(w3|point1)\n",
    "    print(\"p(class_w3|given_point):\",p_of_class_w3_given_point)\n",
    "\n",
    "    if(p_of_class_w1_given_point>p_of_class_w2_given_point and p_of_class_w1_given_point>p_of_class_w3_given_point):\n",
    "        print('given point belongs to class w1')\n",
    "    elif (p_of_class_w2_given_point>p_of_class_w1_given_point and p_of_class_w2_given_point>p_of_class_w3_given_point):\n",
    "        print('given point belongs to class w2')\n",
    "    else:\n",
    "        print('given point belongs to class w3')\n",
    "    \n",
    "    return"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**1. point1 = (0.50, 1.0, 0.0) h=1**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "p(class_w1|given_point): 1.0\n",
      "p(class_w2|given_point): 2.3681437764036732\n",
      "p(class_w3|given_point): 1.0650270356994131\n",
      "given point belongs to class w2\n"
     ]
    }
   ],
   "source": [
    "point1 = np.array([0.5,1,0])\n",
    "Parson_classifier(point1,X_W1,X_W2,X_W3,X_W,h)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**2. point2 = (0.31, 1.51, − 0.50) h=1**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "p(class_w1|given_point): 1.0\n",
      "p(class_w2|given_point): 1.8822840496392448\n",
      "p(class_w3|given_point): 0.9145188828886278\n",
      "given point belongs to class w2\n"
     ]
    }
   ],
   "source": [
    "point2 = np.array([0.31, 1.51, -0.5])\n",
    "Parson_classifier(point2,X_W1,X_W2,X_W3,X_W,h)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**3. point3 = (− 0.3, 0.44, − 0.1) h=1**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "p(class_w1|given_point): 1.0\n",
      "p(class_w2|given_point): 1.361312404003214\n",
      "p(class_w3|given_point): 1.005297656974716\n",
      "given point belongs to class w2\n"
     ]
    }
   ],
   "source": [
    "point3 = np.array([-0.3, 0.44, -0.1])\n",
    "Parson_classifier(point3,X_W1,X_W2,X_W3,X_W,h)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### b) Repeat (a) with h = 0.1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "h = 0.1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**1. point1 = (0.50, 1.0, 0.0) h=0.1**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "p(class_w1|given_point): 1.0\n",
      "p(class_w2|given_point): 771185458094198.9\n",
      "p(class_w3|given_point): 8.598252081943133e+17\n",
      "given point belongs to class w3\n"
     ]
    }
   ],
   "source": [
    "point1 = np.array([0.5,1,0])\n",
    "Parson_classifier(point1,X_W1,X_W2,X_W3,X_W,h)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**2. point2 = (0.31, 1.51, − 0.50) h=0.1**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "p(class_w1|given_point): 1.0\n",
      "p(class_w2|given_point): 9197474026983.35\n",
      "p(class_w3|given_point): 4.091754467860185e+18\n",
      "given point belongs to class w3\n"
     ]
    }
   ],
   "source": [
    "point2 = np.array([0.31, 1.51, -0.5])\n",
    "Parson_classifier(point2,X_W1,X_W2,X_W3,X_W,h)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**3. point3 = (−0.3, 0.44, − 0.1) h=0.1**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "p(class_w1|given_point): 1.0\n",
      "p(class_w2|given_point): 3291.010699244844\n",
      "p(class_w3|given_point): 20743639722.009953\n",
      "given point belongs to class w3\n"
     ]
    }
   ],
   "source": [
    "point3 = np.array([-0.3, 0.44, -0.1])\n",
    "Parson_classifier(point3,X_W1,X_W2,X_W3,X_W,h)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Testing above code on training set"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Testing on point which belongs to class w1 in given training set:**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "p(class_w1|given_point): 1.0\n",
      "p(class_w2|given_point): 0.0\n",
      "p(class_w3|given_point): 0.0007544704218948585\n",
      "given point belongs to class w1\n"
     ]
    }
   ],
   "source": [
    "point = np.array([0.28,1.31,-6.2])\n",
    "Parson_classifier(point,X_W1,X_W2,X_W3,X_W,h)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Testing on point which belongs to class w2 in given training set:**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "p(class_w1|given_point): 1.0\n",
      "p(class_w2|given_point): 337236398295.73505\n",
      "p(class_w3|given_point): 254434887.7004858\n",
      "given point belongs to class w2\n"
     ]
    }
   ],
   "source": [
    "point = np.array([0.011,1.03,-0.21])\n",
    "Parson_classifier(point,X_W1,X_W2,X_W3,X_W,h)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Testing on point which belongs to class w3 in given training set:**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "p(class_w1|given_point): 1.0\n",
      "p(class_w2|given_point): 6.842548550247955e+51\n",
      "p(class_w3|given_point): 1.4694517611204948e+66\n",
      "given point belongs to class w3\n"
     ]
    }
   ],
   "source": [
    "point = np.array([1.36,2.17,0.14])\n",
    "Parson_classifier(point,X_W1,X_W2,X_W3,X_W,h)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}