Tensorflow实现逻辑回归

1. 代码实现：

import numpy as np;
import matplotlib.pyplot as plt;
import tensorflow as tf;
DATA_FILE1 = "./data1.txt";
class LogicRegression:
    def sigmoid(self,z):
        return 1/(1+np.exp(-z));
    def __init__(self):
        data = np.loadtxt(DATA_FILE1, delimiter=",");
        self.x = data[:,0:2].astype(np.float32);
        self.y = data[:,2].astype(np.float32);
        self.pos = np.where(self.y==1);
        self.neg = np.where(self.y==0);
        # rememeber reshape to 1 dimension
        self.x = np.reshape(self.x,newshape=(len(self.y),2));
        self.y = np.reshape(self.y,newshape=(len(self.y),1));
        
    def train(self):
        x = tf.placeholder(tf.float32);
        y = tf.placeholder(tf.float32);
        #w = tf.Variable(np.random.randn(2,1).astype(np.float32));

        #here we need to understand why initialize with zeros
        w = tf.Variable(tf.zeros([2,1]));
        b = tf.Variable(np.random.rand());

        z = tf.matmul(x,w) + b;
        h = (y * tf.log(tf.sigmoid(z)) + (1-y)* tf.log(1-tf.sigmoid(z))) * -1;
        loss = tf.reduce_mean(h);

        optimizer = tf.train.GradientDescentOptimizer(0.002).minimize(loss);
        init = tf.global_variables_initializer();
        
        with tf.Session() as sess:
            sess.run(init);
            for i in range(1000000):
                
                feed = {x:self.x,y:self.y};
                sess.run(optimizer,feed_dict = feed);
                #'''
                if (i % 50000 == 0):
                    #w = sess.run(w);
                    print (sess.run(w).flatten(),sess.run(b).flatten());
                    print ("loss:",sess.run(loss,{x:self.x,y:self.y}));
            self.w = sess.run(w).flatten();
            self.b = sess.run(b).flatten();
            #'''
    def show(self):
        plt.figure();
        plt.scatter(self.x[self.pos,0],self.x[self.pos,1],c='g',marker='o');
        plt.scatter(self.x[self.neg,0],self.x[self.neg,1],c='r',marker='x')
        x = np.linspace(0, 100, 1000)
        y = []
        for i in x:
            y.append((i * -self.w[0] - self.b) / self.w[1])
            
        plt.plot(x, y)
        plt.show();
if __name__ == "__main__":
    LR = LogicRegression();
    LR.train();
    LR.show();

2. 模型介绍：