代码如下:
#-*- coding:utf-8 _*-
"""
@author:bluesli
@file: linear_regression.py
@time: 2018/06/23
"""
#构造数据
#机器学习中,权重值是随机初始化,b是定量初始化一般初始化为1;
#reduce_mean:均方差;
#采用梯度下降的方法来优化;
#构造训练函数:最小化loss
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
#生成随机的点
point_num = 1000
# 定义一个存储点的列表
vectors_set = []
for i in range(point_num):
x = np.random.normal(0.00,0.55)
y = x*0.1+0.3 +np.random.normal(0.0,0.03)
vectors_set.append([x,y])
# 利用列表推导式生成x和y数据列表
x_data = [v[0] for v in vectors_set]
y_data = [v[1] for v in vectors_set]
# x_data =x_data.reshape(1,1)
# y_data = y_data.reshape(1,1)
plt.scatter(x_data,y_data,c='r')
plt.show()
# input_x = tf.placeholder(tf.float32,[1])
# input_y = tf.placeholder(tf.float32,[1])
#生成-1到1之间的随机数
w = tf.Variable(tf.random_uniform([1],-1,1),name='w')
#对于初始的b一般采用常量1来表示
b = tf.Variable(tf.zeros([1]),name='b')
# y_predict = tf.add(tf.matmul(input_x,w),b)
#这里如果用tf自带的matmuul方法则要对数据格式进行转换(shape的改变)
y_predict = x_data * w +b
loss = tf.reduce_mean(tf.square(y_predict - y_data),name='loss')
optimizer = tf.train.GradientDescentOptimizer(0.5)
train = optimizer.minimize(loss,name='train')
init_op = tf.global_variables_initializer()
with tf.Session() as sess:
sess.run(init_op)
#打印出初始的w,和,b和loss
print("W =", sess.run(w), "b =", sess.run(b), "loss =", sess.run(loss))
#训练10次
for step in range(10):
#开始训练
sess.run(train)
# 输出训练好的值
print("W =", sess.run(w), "b =", sess.run(b), "loss =", sess.run(loss))
#把生成的图写进temp文件家中
writer = tf.summary.FileWriter("./tmp", sess.graph)
plt.scatter(x_data,y_data)
plt.scatter(x_data,x_data*sess.run(w)+sess.run(b),c='r')
plt.show()
import sys
print(sys.path)
'''