Source address: https://github.com/w1449550206/Linear-regression-with-tensorflow.git
import tensorflow as tf
with tf.variable_scope("lr_model"):
def linear_regression():#定义名字
# 1）准备好数据集：y = 0.8x + 0.7 100个样本
# 特征值X, 目标值y_true
with tf.variable_scope("original_data"):#定义名字
X = tf.random_normal(shape=(100, 1), mean=2, stddev=2, name="original_data_x")
# y_true [100, 1]
# 矩阵运算 X（100， 1）* （1, 1）= y_true(100, 1)
y_true = tf.matmul(X, [[0.8]], name="original_matmul") + 0.7
# 2）建立线性模型：
# y = W·X + b，目标：求出权重W和偏置b
# 3）随机初始化W1和b1
with tf.variable_scope("linear_model"):#定义名字
weights = tf.Variable(initial_value=tf.random_normal(shape=(1, 1)), name="weights")
bias = tf.Variable(initial_value=tf.random_normal(shape=(1, 1)), name="bias")
y_predict = tf.matmul(X, weights, name="model_matmul") + bias
# 4）确定损失函数（预测值与真实值之间的误差）-均方误差
with tf.variable_scope("loss"):#定义名字
error = tf.reduce_mean(tf.square(y_predict - y_true), name="error_op")
# 5）梯度下降优化损失：需要指定学习率（超参数）
# W2 = W1 - 学习率*(方向)
# b2 = b1 - 学习率*(方向)
with tf.variable_scope("gd_optimizer"):#定义名字
optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.01, name="optimizer").minimize(error)
# 2）收集变量
tf.summary.scalar("error", error)##生成准确率标量图
tf.summary.histogram("weights", weights)#用来显示直方图信息
tf.summary.histogram("bias", bias)#用来显示直方图信息
# 3）合并变量
merge = tf.summary.merge_all()
# 初始化变量
init = tf.global_variables_initializer()
# 开启会话进行训练
with tf.Session() as sess:
# 运行初始化变量Op
sess.run(init)
print("随机初始化的权重为%f， 偏置为%f" % (weights.eval(), bias.eval()))
# 1）创建事件文件【重要】
file_writer = tf.summary.FileWriter(logdir="./summary", graph=sess.graph)
# 训练模型
for i in range(100):
sess.run(optimizer)
print("第%d步的误差为%f，权重为%f， 偏置为%f" % (i, error.eval(), weights.eval(), bias.eval()))
# 4）运行合并变量op
summary = sess.run(merge)
file_writer.add_summary(summary, i)
return None
if __name__ == '__main__':
linear_regression()
Visualization:
