TF0004、实现简单线性回归

import tensorflow as tf
import matplotlib.pyplot as plt
import numpy as np
import warnings
warnings.filterwarnings('ignore')

#构造数据
np.random.seed(1)
X_data = np.random.rand(200)
Y_data = X_data * 0.5 + 2
plt.scatter(X_data,Y_data)
plt.show()

#添加噪声
noise = np.random.normal(0,0.05,X_data.shape)
Y_data = Y_data + noise
plt.scatter(X_data,Y_data)
plt.show()

#构造线性模型
a = tf.Variable(np.random.rand(1))
b = tf.Variable(np.random.rand(1))
y = a*X_data + b
#二次代价函数（损失函数） 均方值
loss = tf.losses.mean_squared_error(Y_data,y)
#设置学习率
lr = 0.2
#定义梯度下降优化器
optimizer = tf.train.GradientDescentOptimizer(lr)
#最小化损失函数
train = optimizer.minimize(loss)

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

with tf.Session() as session:
    session.run(init)
    for i in range(100):
        session.run(train)
        if i%10 == 0 :
            print("第%i次训练[a,b]的值为："%i,session.run([a,b]))
    predict_y = session.run(y)
    
    plt.scatter(X_data,Y_data)
    plt.plot(X_data,predict_y,color='r')
    plt.show()