# 引入包
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt

#准备好数据，x为生成-1到1之间的100个数，y为2*x+1+噪声
data_x = np.linspace(-1, 1, 100)
data_y = np.multiply(data_x, 2) + 1 + np.random.uniform(-0.5, 0.5, 100)

#画图，将x,y的数据投上去
fig = plt.figure()
ax = fig.add_subplot(1,1,1)
ax.scatter(data_x, data_y)
plt.ion()
plt.show()

#准备好placeholder,占位符，也叫容器，也就是个坑，用来放数据的
x_placeholder = tf.placeholder(tf.float32, name='X')
y_placeholder = tf.placeholder(tf.float32, name='Y')

#变量，
W = tf.Variable(dtype=tf.float32,name='W',initial_value=tf.random_normal([1]))
b = tf.Variable(dtype=tf.float32,name='W',initial_value=tf.random_normal([1]))

#y_pred 给出回归式的Op
y_pred = tf.add(tf.multiply(x_placeholder,W),b)

#求loss,给出优化器
loss = tf.reduce_mean(tf.square(y_placeholder-y_pred,name='loss'))

optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.1).minimize(loss)

#初始化变量
init = tf.global_variables_initializer()

#session会话
with tf.Session() as sess:
    sess.run(init)
	#迭代100次
    for i in range(100):
        total_loss = 0
        for x, y in zip(data_x,data_y ):
			#将之前graph中的loss和optimizer跑起来，训练过程就在这，因为有个minimize(loss)的操作	
            _, los = sess.run([optimizer, loss], feed_dict={x_placeholder: x, y_placeholder: y})
            total_loss += los
		#求预测值
        yp = sess.run(y_pred,feed_dict={x_placeholder:data_x, y_placeholder: data_y})
        if i % 5 == 0:
			#每5步画一次线
            lines = ax.plot(data_x, yp, 'r-', lw=5)
            print('Epoch {0}: {1}'.format(i, total_loss / 100))
            plt.pause(0.1)
            try:
			#为了不让线一直赖着不走，我们每画完就要删除。
                ax.lines.pop(0)
            except Exception:
                print('error')