minibatch.py

import tensorflow as tf

(x_train, y_train), (x_test, y_test) = tf.keras.datasets.boston_housing.load_data()

x_train_mean = x_train.mean(axis=0)
x_train_std = x_train.std(axis=0)
y_train_mean = y_train.mean()
y_train_std = y_train.std()

x_train = (x_train-x_train_mean)/x_train_std
y_train = (y_train-y_train_mean)/y_train_std

x_test_mean = x_test.mean(axis=0)
x_test_std = x_test.std(axis=0)
y_test_mean = y_test.mean()
y_test_std = y_test.std()

x_test = (x_test-x_test_mean)/x_test_std
y_test = (y_test-y_test_mean)/y_test_std

x = tf.placeholder(tf.float32, (None, 13), name='x')
y = tf.placeholder(tf.float32, (None, 1), name='y')

w = tf.Variable(tf.random_normal((13,1)))
pred = tf.matmul(x, w)

loss = tf.reduce_mean((y-pred)**2)
optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.1)
train_step = optimizer.minimize(loss)

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    for step in range(100):
        train_loss, _ = sess.run(
            [loss, train_step],
            feed_dict={
                x: x_train,
                y: y_train.reshape((-1,1))
            }
        )
        print('step:{}, train_loss:{}'.format(step, train_loss))
        
    pred_ = sess.run(
        pred,
        feed_dict={x:x_test}
    )
    
    
import numpy as np

def get_batches(x, y, batch_size, shuffle=True):
    n_data = len(x)
    indices = np.arange(n_data)
    if shuffle:
        np.random.shuffle(indices)
    x_shuffled = x[indices]
    y_shuffled = y[indices]
    
    for i in range(0, n_data, batch_size):
        x_batch = x_shuffled[i: i+batch_size]
        y_batch = y_shuffled[i: i+batch_size]
        
        yield x_batch, y_batch
        
        
BATCH_SIZE = 32

step = 0
with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    for epoch in range(100):
        for x_batch, y_batch in get_batches(x_train, y_train, 32):   
            train_loss, _ = sess.run(
                [loss, train_step],
                feed_dict={
                    x: x_batch,
                    y: y_batch.reshape((-1,1))
                }
            )
            print('step:{}, train_loss:{}'.format(step, train_loss))
            step += 1
        
    pred_ = sess.run(
        pred,
        feed_dict={x:x_test}
    )