Implement_NN.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "A data: [5.1 3.5 1.4 0.2] \n",
      "Label: 0\n"
     ]
    }
   ],
   "source": [
    "from sklearn import datasets\n",
    "iris = datasets.load_iris()\n",
    "X = iris.data # features, 'sepal length (cm)', 'sepal width (cm)', 'petal length (cm)', 'petal width (cm)'\n",
    "y = iris.target # label, 'setosa' 'versicolor' 'virginica'\n",
    "\n",
    "print('A data:', X[0], '\\nLabel:', y[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
      " 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      " 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      " 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      " 0 0]\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "re_y = np.array([1 if i==0 else 0 for i in y]) # 'setosa':1, 'not setosa':0\n",
    "print(re_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "def plot(xfeature_name, yfeature_name, xfeature, yfeature, ori_label, re_label):\n",
    "    # setting color\n",
    "    ori_color = {0: 'red', 1: 'blue', 2: 'yellow'} # label, 0:'setosa' 1:'versicolor' 2:'virginica'\n",
    "    relbl_color = {0: 'blue', 1: 'red'} # relabel, 0:'not setosa' 1:'setosa'\n",
    "    # plot\n",
    "    plt.figure()\n",
    "    plt.subplot(1, 2, 1)\n",
    "    plt.subplots_adjust(wspace = 0.5)\n",
    "    plt.title(\"Original Label\")\n",
    "    plt.xlabel(xfeature_name)\n",
    "    plt.ylabel(yfeature_name)\n",
    "    for i, j, color in zip(xfeature, yfeature, ori_label):\n",
    "        plt.scatter(i, j, c=ori_color[color], marker='o', s=50, edgecolor='k',cmap=plt.cm.Spectral)\n",
    "\n",
    "    plt.subplot(1, 2, 2)\n",
    "    plt.title(\"Relabel\")\n",
    "    plt.xlabel(xfeature_name)\n",
    "    plt.ylabel(yfeature_name)\n",
    "    for i, j, color in zip(xfeature, yfeature, re_label):\n",
    "        plt.scatter(i, j, c=relbl_color[color], marker='o', s=50, edgecolor='k',cmap=plt.cm.Spectral)\n",
    "\n",
    "\n",
    "plot('sepal length(cm)', 'sepal width(cm)', X[:, 0], X[:, 1], y, re_y)\n",
    "plot('petal length(cm)', 'petal width(cm)', X[:, 2], X[:, 3], y, re_y)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "Desing ANN Model\n",
    "\"\"\"\n",
    "class my_NN(object):\n",
    "    def __init__(self, input_units, output_units, hidden_units):\n",
    "        self.input = input_units # feature numbers\n",
    "        self.output = output_units # class number \n",
    "        self.hidden_units = hidden_units # single layer\n",
    "        \n",
    "        # initialize matrix of weights;\n",
    "        np.random.seed(1) \n",
    "        # weight1: input -> hidden layer \n",
    "        self.w1 = np.random.randn(self.input, self.hidden_units)\n",
    "        self.b1 = np.zeros(shape=(self.hidden_units, 1))        \n",
    "\n",
    "        # weight2: hidden layer -> output\n",
    "        self.w2 = np.random.randn(self.hidden_units, self.output)\n",
    "        self.b2 = np.zeros(shape=(self.output, 1))\n",
    "        \n",
    "\n",
    "    def _forward_propagation(self, X):\n",
    "        # W^T dot X\n",
    "        self.z2 = np.dot(self.w1.T, X.T) + self.b1\n",
    "        self.a2 = self._sigmoid(self.z2)\n",
    "        self.z3 = np.dot(self.w2.T, self.a2) + self.b2\n",
    "        self.a3 = self._sigmoid(self.z3)\n",
    "        return self.a3\n",
    "\n",
    "    def _sigmoid(self, z):\n",
    "        return 1/(1+np.exp(-z))\n",
    "    \n",
    "    def _sigmoid_prime(self, z):\n",
    "        return self._sigmoid(z)*(1-self._sigmoid(z))\n",
    "    \n",
    "    def _backward_propagation(self, X, y):\n",
    "        predict = self._forward_propagation(X)\n",
    "        m = X.shape[0]\n",
    "        delta3 = predict - y\n",
    "        dz3 = np.multiply(delta3, self._sigmoid_prime(self.z3))\n",
    "        self.dw2 = (1/m)*np.sum(np.multiply(self.a2, dz3), axis=1).reshape(self.w2.shape)\n",
    "        \n",
    "        delta2 = delta3*self.w2*self._sigmoid_prime(self.z2)\n",
    "        self.dw1 = (1/m)*np.dot(X.T, delta2.T)\n",
    "\n",
    "    def _update(self, learning_rate=1.2):\n",
    "        self.w1 = self.w1 - learning_rate*self.dw1\n",
    "        self.w2 = self.w2 - learning_rate*self.dw2\n",
    "        \n",
    "    def _loss(self, predict, y):\n",
    "        logprobs = np.multiply(np.log(predict), y) + np.multiply((1 - y), np.log(1 - predict))\n",
    "        loss = - np.sum(logprobs) / y.shape[0]\n",
    "        return loss\n",
    "    \n",
    "    def predict(self, X):\n",
    "        y_hat = self._forward_propagation(X)\n",
    "        y_hat = [1 if i[0]>= 0.5 else 0 for i in y_hat.T]\n",
    "        return np.array(y_hat)\n",
    "    \n",
    "    def train(self, X, y, iteration=34):\n",
    "        for i in range(iteration):\n",
    "            y_hat = self._forward_propagation(X)\n",
    "            loss = self._loss(y_hat, y)\n",
    "            self._backward_propagation(X,y)\n",
    "            self._update()\n",
    "            if i%10==0:\n",
    "                print(\"loss: \", loss)\n",
    "                \n",
    "    def score(self, predict, y):\n",
    "        cnt = np.sum(predict==y)\n",
    "        return (cnt/len(y))*100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss:  0.8397954104431425\n",
      "loss:  0.527004500613035\n",
      "loss:  0.45066235073937466\n",
      "loss:  0.3951326953572771\n",
      "predcit:  [0 1 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0\n",
      " 0]\n",
      "anser: [0 1 0 0 0 0 0 1 1 0 1 1 1 0 0 1 0 1 1 0 1 0 0 1 1 1 1 1 1 0 0 1 0 1 0 0 0\n",
      " 0]\n",
      "score:  71.05263157894737\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "if __name__=='__main__':\n",
    "\n",
    "    train_X, test_X, train_y, test_y = train_test_split(X, re_y, random_state=4, test_size=0.25) # split training data and testing data\n",
    "    input_feature = train_X.shape[1]\n",
    "    output_units = 1\n",
    "    hidden_units = 6\n",
    "    clr = my_NN(input_feature, output_units, hidden_units) #initialize the model\n",
    "    clr.train(train_X, train_y) # train model\n",
    "    pre_y = clr.predict(test_X) # predict\n",
    "    score = clr.score(pre_y, test_y) # get the accuracy score\n",
    "    print('predcit: ', pre_y)\n",
    "    print('anser:', test_y)\n",
    "    print('score: ', score)    "
   ]
  }
 ],
 "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}