一元回归模型记录

使用Tensorflow完成

#加载包
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt

#定义超参数，就是模型训练过程中的参数，包括学习率、隐层神经元个数、批数据个数
learning_rate = 0.01
max_train_steps = 1000
log_step = 5

#训练集
train_X = np.array([[3.3],[4.4],[5.5],[6.71],[6.93],[4.168],[9.779],[6.182],[7.59],[2.167], [7.042],[10.791],[5.313],[7.997],[5.654],[9.27],[3.1]], dtype=np.float32)
train_Y = np.array([[1.7],[2.76],[2.09],[3.19],[1.694],[1.573],[3.366],[2.596],[2.53],  [1.221],[2.827],[3.465],[1.65],[2.904],[2.42],[2.94],[1.3]], dtype=np.float32)
total_samples = train_X.shape[0]

#构建模型
#输入数据
X = tf.placeholder(tf.float32, [None, 1])
#模型参数
W = tf.Variable(tf.random_normal([1,1]), name="weight")
b = tf.Variable(tf.zeros([1]), name="bias")
#推理值
Y = tf.matmul(X, W) + b

#实际值
Y_ = tf.placeholder(tf.float32, [None, 1])
#均方差损失
loss = tf.reduce_sum(tf.pow(Y-Y_, 2)) / (total_samples)
#随机梯度下降
optimizer = tf.train.GradientDescentOptimizer(learning_rate)
#最小化损失值
train_op = optimizer.minimize(loss)

with tf.Session() as sess:
    #初始化全局变量
    sess.run(tf.global_variables_initializer())
    print("Start training:")
    #分布训练
    for step in range(max_train_steps):
        sess.run(train_op, feed_dict={X: train_X, Y_:train_Y})
        #打印日志 每log_step步打印损失、权值和偏置
        if step % log_step == 0:
            c = sess.run(loss, feed_dict={X: train_X, Y_:train_Y})
            print("Step%d:%d, loss==%.4f, W==%.4f, b==%.4f"%(step, max_train_steps, c, sess.run(W), sess.run(b)))
    #结果的损失
    final_loss = sess.run(loss, feed_dict={X: train_X, Y_:train_Y})
    print("Step:%d, loss==%.4f, W==%.4f, b==%.4f"%(max_train_steps, final_loss, sess.run(W), sess.run(b)))
    #结果模型输出
    weight, bias = sess.run([W, b])
    print("Linear Regression Model: Y==%.4f*X+%.4f"%(weight, bias))
    

Start training:
Step0:1000, loss==1.1118, W==0.1942, b==0.1869
Step5:1000, loss==0.1962, W==0.3341, b==0.2140
Step10:1000, loss==0.1952, W==0.3331, b==0.2210
Step15:1000, loss==0.1942, W==0.3321, b==0.2280
Step20:1000, loss==0.1932, W==0.3312, b==0.2349
Step25:1000, loss==0.1923, W==0.3302, b==0.2417
Step30:1000, loss==0.1914, W==0.3293, b==0.2484
Step35:1000, loss==0.1905, W==0.3283, b==0.2551
......
Step965:1000, loss==0.1543, W==0.2596, b==0.7421
Step970:1000, loss==0.1542, W==0.2595, b==0.7428
Step975:1000, loss==0.1542, W==0.2594, b==0.7435
Step980:1000, loss==0.1542, W==0.2593, b==0.7441
Step985:1000, loss==0.1542, W==0.2593, b==0.7448
Step990:1000, loss==0.1542, W==0.2592, b==0.7454
Step995:1000, loss==0.1542, W==0.2591, b==0.7461
Step:1000, loss==0.1542, W==0.2590, b==0.7466
Linear Regression Model: Y==0.2590*X+0.7466

#可视化
%matplotlib inline
plt.plot(train_X, train_Y, 'ro', label='Training data')
plt.plot(train_X, weight * train_X + bias, label='Fitted line')
#添加图例
plt.legend()
#显示图像
plt.show()

output_5_0.png