tf_linear_regression_part2.py

import tensorflow as tf
import numpy

# Parameters
learning_rate = 0.01
training_epochs = 500
datapoints_count = 100

# Training Data
train_X = []
train_Y = []
for i in range(datapoints_count):
  train_X.append(i)
  train_Y.append(5*i+8.0)

train_X = numpy.asarray(train_X)
train_Y = numpy.asarray(train_Y)
n_samples = train_X.shape[0]

# tf Graph Input
X = tf.placeholder("float", name="X")
Y = tf.placeholder("float", name="Y")

# Set model weights
W = tf.Variable(0.0, name="weight")
tf.summary.scalar('W', W)
b = tf.Variable(0.0, name="bias")
tf.summary.scalar('b', b)

# Construct a linear model
y_pred = X * W + b

# Mean squared error
cost = tf.div(tf.reduce_sum(tf.square(y_pred-Y)), (2*n_samples), name="cost")

# Gradient descent
optimizer = tf.train.GradientDescentOptimizer(learning_rate).minimize(cost)

saver = tf.train.Saver()

# Start training
with tf.Session() as sess:
  merged = tf.summary.merge_all()
  train_writer = tf.summary.FileWriter('./train', sess.graph)

  sess.run(tf.global_variables_initializer())

  # Fit all training data
  for epoch in range(training_epochs):
    for (x, y) in zip(train_X, train_Y):
      sess.run(optimizer, feed_dict={X: x, Y: y})

    training_cost, summary = sess.run([cost, merged], feed_dict={X: train_X, Y: train_Y})
    train_writer.add_summary(summary, epoch)

    # Display logs for every 50 epoch step
    if (epoch+1) % 50 == 0:
      print("Epoch: %04d cost=%.9f W=%f b=%f" % \
        ((epoch+1), training_cost, sess.run(W), sess.run(b)))

  print("Training cost=%.9f W=%f b=%f" % \
    (training_cost, sess.run(W), sess.run(b)))

  saved_path = saver.save(sess, './saved_variable')
  print('model saved in {}'.format(saved_path))