Perceptron.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "5a3c90f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "880f2fa1",
   "metadata": {},
   "outputs": [],
   "source": [
    "class Percepton:\n",
    "    \n",
    "    def __init__(self, eta: float=None, epochs: int=None):\n",
    "        self.weights = np.random.randn(3) * 1e-4 # small values\n",
    "        training = eta is not None and epochs is not None\n",
    "        if training:\n",
    "            self.eta = eta \n",
    "            self.epochs = epochs\n",
    "            \n",
    "    def _z_outcome(self, x_with_bias, weights):\n",
    "        return np.dot(x_with_bias, weights)\n",
    "    \n",
    "    def activation_function(self, z):\n",
    "        return np.where(z > 0, 1, 0)\n",
    "    \n",
    "    def fit(self, X, y):\n",
    "        self.X_train = X\n",
    "        self.Y_train = y\n",
    "            \n",
    "        X_with_bias = np.c_[self.X_train, -np.ones((len(self.X_train), 1))]\n",
    "        print('X_with bias: ', X_with_bias)\n",
    "        \n",
    "        for epoch in range(self.epochs):\n",
    "            print('--'*10)\n",
    "            print('for epoch >> {0}'.format(epoch))\n",
    "            print('--'*10)\n",
    "            \n",
    "            z = self._z_outcome(X_with_bias, self.weights)\n",
    "            y_hat = self.activation_function(z)\n",
    "            print(f\"predicted value after forward pass: \\n{y_hat}\")\n",
    "            \n",
    "            self.error = y - y_hat\n",
    "            print('Error -->', self.error)\n",
    "            \n",
    "            \n",
    "            self.weights = self.weights + self.eta * np.dot(X_with_bias.T, self.error)\n",
    "            print(f\"updated weights after epoch: {epoch + 1}/{self.epochs}: \\n{self.weights}\")\n",
    "            print(\"##\"*10)\n",
    "            \n",
    "    \n",
    "    def predict(self, X_test):\n",
    "        X_with_bias = np.c_[X_test, -np.ones((len(X_test), 1))]\n",
    "        z = self._z_outcome(X_with_bias)\n",
    "        return self.activation_function(z)\n",
    "    \n",
    "    def total_loss(self):\n",
    "        total_loss = np.sum(self.error)\n",
    "        print(f\"\\ntotal loss: {total_loss}\\n\")\n",
    "        return total_loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "f557a13f",
   "metadata": {},
   "outputs": [],
   "source": [
    "def return_x_y(df):\n",
    "    X = df.drop('y', axis=1)\n",
    "    y = df.y\n",
    "    return X, y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "e4a45c52",
   "metadata": {},
   "outputs": [],
   "source": [
    "AND = {\n",
    "    \"x1\": [0,0,1,1],\n",
    "    \"x2\": [0,1,0,1],\n",
    "    \"y\" : [0,0,0,1]\n",
    "}\n",
    "\n",
    "df_and = pd.DataFrame(AND)\n",
    "X, y = return_x_y(df_and)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "99c26931",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(   x1  x2\n",
       " 0   0   0\n",
       " 1   0   1\n",
       " 2   1   0\n",
       " 3   1   1,\n",
       " 0    0\n",
       " 1    0\n",
       " 2    0\n",
       " 3    1\n",
       " Name: y, dtype: int64)"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X, y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "1c8034f0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_with bias:  [[ 0.  0. -1.]\n",
      " [ 0.  1. -1.]\n",
      " [ 1.  0. -1.]\n",
      " [ 1.  1. -1.]]\n",
      "--------------------\n",
      "for epoch >> 0\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[1 1 1 0]\n",
      "Error --> 0   -1\n",
      "1   -1\n",
      "2   -1\n",
      "3    1\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 1/10: \n",
      "[-4.48861482e-05 -7.08974477e-05  1.99890154e-01]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 1\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 0 0 0]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    1\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 2/10: \n",
      "[0.09995511 0.0999291  0.09989015]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 2\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 1 1 1]\n",
      "Error --> 0    0\n",
      "1   -1\n",
      "2   -1\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 3/10: \n",
      "[-4.48861482e-05 -7.08974477e-05  2.99890154e-01]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 3\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 0 0 0]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    1\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 4/10: \n",
      "[0.09995511 0.0999291  0.19989015]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 4\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 0 0 0]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    1\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 5/10: \n",
      "[0.19995511 0.1999291  0.09989015]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 5\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 1 1 1]\n",
      "Error --> 0    0\n",
      "1   -1\n",
      "2   -1\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 6/10: \n",
      "[0.09995511 0.0999291  0.29989015]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 6\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 0 0 0]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    1\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 7/10: \n",
      "[0.19995511 0.1999291  0.19989015]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 7\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 1 1 1]\n",
      "Error --> 0    0\n",
      "1   -1\n",
      "2   -1\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 8/10: \n",
      "[0.09995511 0.0999291  0.39989015]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 8\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 0 0 0]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    1\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 9/10: \n",
      "[0.19995511 0.1999291  0.29989015]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 9\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 0 0 1]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 10/10: \n",
      "[0.19995511 0.1999291  0.29989015]\n",
      "####################\n",
      "\n",
      "total loss: 0\n",
      "\n"
     ]
    }
   ],
   "source": [
    "X, y = prepare_data(df_AND)\n",
    "\n",
    "ETA = 0.1 # 0 and 1\n",
    "EPOCHS = 10\n",
    "\n",
    "model_and = Percepton(eta=ETA, epochs=EPOCHS)\n",
    "model_and.fit(X, y)\n",
    "\n",
    "_ = model_and.total_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "915c85d5",
   "metadata": {},
   "outputs": [],
   "source": [
    "OR = {\n",
    "    \"x1\": [0,0,1,1],\n",
    "    \"x2\": [0,1,0,1],\n",
    "    \"y\" : [0,1,1,1]\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "83a5dc03",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_or = pd.DataFrame(OR)\n",
    "X_or, Y_or = return_x_y(df_or)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "66dc7e16",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_with bias:  [[ 0.  0. -1.]\n",
      " [ 0.  1. -1.]\n",
      " [ 1.  0. -1.]\n",
      " [ 1.  1. -1.]]\n",
      "--------------------\n",
      "for epoch >> 0\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 0 1 0]\n",
      "Error --> 0    0\n",
      "1    1\n",
      "2    0\n",
      "3    1\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 1/10: \n",
      "[ 0.10005872  0.19986413 -0.19996165]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 1\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[1 1 1 1]\n",
      "Error --> 0   -1\n",
      "1    0\n",
      "2    0\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 2/10: \n",
      "[ 0.10005872  0.19986413 -0.09996165]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 2\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[1 1 1 1]\n",
      "Error --> 0   -1\n",
      "1    0\n",
      "2    0\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 3/10: \n",
      "[1.00058720e-01 1.99864130e-01 3.83468014e-05]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 3\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 1 1 1]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 4/10: \n",
      "[1.00058720e-01 1.99864130e-01 3.83468014e-05]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 4\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 1 1 1]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 5/10: \n",
      "[1.00058720e-01 1.99864130e-01 3.83468014e-05]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 5\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 1 1 1]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 6/10: \n",
      "[1.00058720e-01 1.99864130e-01 3.83468014e-05]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 6\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 1 1 1]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 7/10: \n",
      "[1.00058720e-01 1.99864130e-01 3.83468014e-05]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 7\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 1 1 1]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 8/10: \n",
      "[1.00058720e-01 1.99864130e-01 3.83468014e-05]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 8\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 1 1 1]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 9/10: \n",
      "[1.00058720e-01 1.99864130e-01 3.83468014e-05]\n",
      "####################\n",
      "--------------------\n",
      "for epoch >> 9\n",
      "--------------------\n",
      "predicted value after forward pass: \n",
      "[0 1 1 1]\n",
      "Error --> 0    0\n",
      "1    0\n",
      "2    0\n",
      "3    0\n",
      "Name: y, dtype: int64\n",
      "updated weights after epoch: 10/10: \n",
      "[1.00058720e-01 1.99864130e-01 3.83468014e-05]\n",
      "####################\n"
     ]
    }
   ],
   "source": [
    "percepton_or = Percepton(eta=0.1, epochs=10)\n",
    "percepton_or.fit(X_or, Y_or)"
   ]
  }
 ],
 "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": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}