Artificial Neural Network Regression Model .ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np \n",
    "import matplotlib.pyplot as plt \n",
    "import pandas as pd \n",
    "from sklearn.metrics import classification_report, accuracy_score, confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def sigmoid(x):\n",
    "    return 1/(1+np.exp(-x))\n",
    "def sigmoid_der(x):\n",
    "    return sigmoid(x)*(1-sigmoid(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training_Set_X :  [[300.    99.     6.8 ]\n",
      " [308.   103.     8.36]\n",
      " [329.   110.     9.15]\n",
      " [332.   118.     9.36]]\n",
      "Training_Set_Y :  [0.36 0.7  0.84 0.9 ]\n",
      "X_test :  [[296.    95.     7.54]\n",
      " [293.    97.     7.8 ]\n",
      " [325.   112.     8.96]]\n",
      "Y_test :  [0.44 0.64 0.8 ]\n",
      "lr:  0.18182496720710062 \n",
      "\n",
      "weights1: [[0.37454012 0.95071431 0.73199394]\n",
      " [0.59865848 0.15601864 0.15599452]\n",
      " [0.05808361 0.86617615 0.60111501]] \n",
      "\n",
      "weights2: [0.70807258 0.02058449 0.96990985] \n",
      "\n"
     ]
    }
   ],
   "source": [
    "Training_Set_X=np.array([[300.0,99.0,6.8],[308.0,103.0,8.36],[329.0,110.0,9.15],[332.0,118.0,9.36]])\n",
    "X_test=np.array([[296.0,95.0,7.54],[293.0,97.0,7.8],[325.0, 112.0,8.96]])\n",
    "Training_Set_Y=np.array([0.36,0.7,0.84,0.9])\n",
    "Y_test=np.array([0.44,0.64,0.8])\n",
    "\n",
    "\n",
    "print(\"Training_Set_X : \",Training_Set_X)\n",
    "print(\"Training_Set_Y : \",Training_Set_Y)\n",
    "print(\"X_test : \",X_test)\n",
    "print(\"Y_test : \",Y_test)\n",
    "\n",
    "#2.\tInitialize the parameters with some random values.\n",
    "np.random.seed(42)\n",
    "weights1 = np.random.rand(3,3)\n",
    "weights2=np.random.rand(3,)\n",
    "bias1 = np.random.rand(1)\n",
    "bias2 = np.random.rand(1)\n",
    "#learning rate\n",
    "lr = np.random.rand()\n",
    "print(\"lr: \",lr,\"\\n\")\n",
    "print(\"weights1:\",weights1,\"\\n\")\n",
    "print(\"weights2:\",weights2,\"\\n\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "At Epoch:  0\n",
      "R : [212.29275343 196.93659604 108.0965468 ]\n",
      "H1 ,H2,& H3 : [1. 1. 1.]\n",
      "S : [1.91090603]\n",
      "Yhat : [0.8711209]\n",
      "Error : [0.5111209]\n",
      "The content of the weight2 matrices  :\n",
      " [0.69763888 0.0101508  0.95947616] \n",
      "\n",
      "The content of the weight1 matrices  :\n",
      " [[0.37454012 0.95071431 0.73199394]\n",
      " [0.59865848 0.15601864 0.15599452]\n",
      " [0.05808361 0.86617615 0.60111501]] \n",
      "\n",
      "At Epoch:  1\n",
      "R : [220.23384216 202.59328993 112.9636597 ]\n",
      "H1 ,H2,& H3 : [1. 1. 1.]\n",
      "S : [1.87960495]\n",
      "Yhat : [0.86756574]\n",
      "Error : [0.16756574]\n",
      "The content of the weight2 matrices  :\n",
      " [0.69413829 0.00665021 0.95597557] \n",
      "\n",
      "The content of the weight1 matrices  :\n",
      " [[0.37454012 0.95071431 0.73199394]\n",
      " [0.59865848 0.15601864 0.15599452]\n",
      " [0.05808361 0.86617615 0.60111501]] \n",
      "\n",
      "At Epoch:  2\n",
      "R : [235.33246001 216.38048425 120.72152944]\n",
      "H1 ,H2,& H3 : [1. 1. 1.]\n",
      "S : [1.86910318]\n",
      "Yhat : [0.86635447]\n",
      "Error : [0.02635447]\n",
      "The content of the weight2 matrices  :\n",
      " [0.69358346 0.00609538 0.95542074] \n",
      "\n",
      "The content of the weight1 matrices  :\n",
      " [[0.37454012 0.95071431 0.73199394]\n",
      " [0.59865848 0.15601864 0.15599452]\n",
      " [0.05808361 0.86617615 0.60111501]] \n",
      "\n",
      "At Epoch:  3\n",
      "R : [244.21551355 219.45736768 127.95142359]\n",
      "H1 ,H2,& H3 : [1. 1. 1.]\n",
      "S : [1.8674387]\n",
      "Yhat : [0.86616163]\n",
      "Error : [-0.03383837]\n",
      "The content of the weight2 matrices  :\n",
      " [0.69429672 0.00680863 0.95613399] \n",
      "\n",
      "The content of the weight1 matrices  :\n",
      " [[0.37454012 0.95071431 0.73199394]\n",
      " [0.59865848 0.15601864 0.15599452]\n",
      " [0.05808361 0.86617615 0.60111501]] \n",
      "\n",
      "SSE: [0.29116245]\n"
     ]
    }
   ],
   "source": [
    "# SE to save all Squared Errors values\n",
    "SE=list()\n",
    "\n",
    "#3.Perform a feed-forward stage to propagate the input forward through the network. \n",
    "for epoch in range(len(Training_Set_X)):\n",
    "    print(\"At Epoch: \",epoch)\n",
    " # feedforward step1 \n",
    "    R = Training_Set_X[epoch].dot( weights1.T) + bias1 # linear response\n",
    "    #feedforward step2 \n",
    "    H = sigmoid(R)  # activation f’n\n",
    "    S=H.dot( weights2.T)+bias2  # linear response\n",
    "    # backpropagation step 1 \n",
    "    Yhat=sigmoid(S)\n",
    "    error = (Yhat) - Training_Set_Y[epoch]\n",
    "    SE.append((error**2))\n",
    "\n",
    "#5.\tPrint the content of the input/output of hidden layers and output layer at the feed forward stages.\n",
    "  #input of hidden layers\n",
    "    print(\"R :\",R)\n",
    "    #the output of hidden layers\n",
    "    print(\"H1 ,H2,& H3 :\" ,H)\n",
    "   #input of output layer\n",
    "    print(\"S :\",S)\n",
    "    #the output of output layer\n",
    "    print(\"Yhat :\",Yhat)\n",
    "    print(\"Error :\",error)\n",
    "    \n",
    "    #4.Preform the back-propagation algorithm using sum square error (SSE).\n",
    "    # backpropagation step 2\n",
    "    S_Si_der = sigmoid_der(S)\n",
    "    G2= (error * S_Si_der)*(H)\n",
    "    R_Si_der=sigmoid_der(R)\n",
    "   # Update weights2    \n",
    "   #6.Print the content of the weight matrices after each epoch (4 epochs)\n",
    "    for i in range(len(weights2)):\n",
    "        weights2[i] -= (lr * G2[i])\n",
    "    print(\"The content of the weight2 matrices  :\\n\",weights2,\"\\n\")\n",
    "        # Update weights1   \n",
    "    for i in range(len(weights1)):        \n",
    "        for j in range(len(weights1[i])):\n",
    "            weights1[i][j] -= (error * S_Si_der*weights2[i]*R_Si_der[i]*Training_Set_X[epoch][j])\n",
    "           \n",
    "    print(\"The content of the weight1 matrices  :\\n\",weights1,\"\\n\")\n",
    "#sum square error \n",
    "print(\"SSE:\",sum(SE))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Example  : 1\n",
      "The predicted chance of acceptance : [0.86640949]\n",
      "Error : [0.42640949]\n",
      "Example  : 2\n",
      "The predicted chance of acceptance : [0.86640949]\n",
      "Error : [0.22640949]\n",
      "Example  : 3\n",
      "The predicted chance of acceptance : [0.86640949]\n",
      "Error : [0.06640949]\n",
      "SSE: [0.23749653]\n"
     ]
    }
   ],
   "source": [
    "#7.\tTest and report the result of classifying the testing examples.\n",
    "# SE2 to save all Squared Errors values\n",
    "SE2=list()\n",
    "for epoch in range(len(X_test)):\n",
    "    # feedforward step1 \n",
    "    R = X_test[epoch].dot( weights1.T) + bias1 # linear response\n",
    "    #feedforward step2 \n",
    "    H = sigmoid(R)  # activation f’n\n",
    "    S=H.dot( weights2.T)+bias2  # linear response\n",
    "    # backpropagation step 1 \n",
    "    Yhat=sigmoid(S)\n",
    "    error = (Yhat) - Y_test[epoch]\n",
    "    SE2.append((error**2))\n",
    "#5.\tPrint the content of the input/output of hidden layers and output layer at the feed forward stages.\n",
    "    print(\"Example  :\",epoch+1)\n",
    "    print(\"The predicted chance of acceptance :\",Yhat)\n",
    "    print(\"Error :\",error)\n",
    "#sum square error \n",
    "print(\"SSE:\",sum(SE2))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}