"""Simple tutorial for using TensorFlow to compute a linear regression.

Parag K. Mital, Jan. 2016"""

'Simple tutorial for using TensorFlow to compute a linear regression.\n\nParag K. Mital, Jan. 2016'

# %% imports
%matplotlib inline
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt

# %% Let's create some toy data
plt.ion()
n_observations = 100
fig, ax = plt.subplots(1, 1)
xs = np.linspace(-3, 3, n_observations)
ys = np.sin(xs) + np.random.uniform(-0.5, 0.5, n_observations)
ax.scatter(xs, ys)
fig.show()
plt.draw()

/home/heythisischo/anaconda2/lib/python2.7/site-packages/matplotlib/figure.py:397: UserWarning: matplotlib is currently using a non-GUI backend, so cannot show the figure
  "matplotlib is currently using a non-GUI backend, "

# %% tf.placeholders for the input and output of the network. Placeholders are
# variables which we need to fill in when we are ready to compute the graph.
X = tf.placeholder(tf.float32)
Y = tf.placeholder(tf.float32)

# %% We will try to optimize min_(W,b) ||(X*w + b) - y||^2
# The `Variable()` constructor requires an initial value for the variable,
# which can be a `Tensor` of any type and shape. The initial value defines the
# type and shape of the variable. After construction, the type and shape of
# the variable are fixed. The value can be changed using one of the assign
# methods.
W = tf.Variable(tf.random_normal([1]), name='weight')
b = tf.Variable(tf.random_normal([1]), name='bias')
Y_pred = tf.add(tf.mul(X, W), b)

# %% Loss function will measure the distance between our observations
# and predictions and average over them.
cost = tf.reduce_sum(tf.pow(Y_pred - Y, 2)) / (n_observations - 1)

# %% if we wanted to add regularization, we could add other terms to the cost,
# e.g. ridge regression has a parameter controlling the amount of shrinkage
# over the norm of activations. the larger the shrinkage, the more robust
# to collinearity.
# cost = tf.add(cost, tf.mul(1e-6, tf.global_norm([W])))

# %% Use gradient descent to optimize W,b
# Performs a single step in the negative gradient
learning_rate = 0.01
optimizer = tf.train.GradientDescentOptimizer(learning_rate).minimize(cost)

# %% We create a session to use the graph
n_epochs = 1000
with tf.Session() as sess:
    # Here we tell tensorflow that we want to initialize all
    # the variables in the graph so we can use them
    sess.run(tf.initialize_all_variables())

    # Fit all training data
    prev_training_cost = 0.0
    for epoch_i in range(n_epochs):
        for (x, y) in zip(xs, ys):
            sess.run(optimizer, feed_dict={X: x, Y: y})

        training_cost = sess.run(
            cost, feed_dict={X: xs, Y: ys})
        print(training_cost)

        if epoch_i % 20 == 0:
            ax.plot(xs, Y_pred.eval(
                feed_dict={X: xs}, session=sess),
                    'k', alpha=epoch_i / n_epochs)
            fig.show()
            plt.draw()

        # Allow the training to quit if we've reached a minimum
        if np.abs(prev_training_cost - training_cost) < 0.000001:
            break
        prev_training_cost = training_cost
fig.show()

1.43989
1.30686
1.18916
1.08503
0.992879
0.911325
0.839141
0.77524
0.718663
0.668564
0.624192
0.584885
0.550058
0.519194
0.491835
0.467578
0.446063
0.426977
0.410039
0.395001
0.381647
0.369783
0.359238
0.349861
0.341519
0.334094
0.327481
0.321587
0.316332
0.311643
0.307455
0.303713
0.300366
0.29737
0.294686
0.292278
0.290117
0.288175
0.286427
0.284853
0.283433
0.282151
0.280991
0.279942
0.27899
0.278125
0.277339
0.276623
0.27597
0.275373
0.274826
0.274325
0.273865
0.273442
0.273052
0.272693
0.27236
0.272052
0.271767
0.271501
0.271254
0.271024
0.27081
0.270609
0.270421
0.270245
0.27008
0.269925
0.269779
0.269642
0.269512
0.26939
0.269274
0.269165
0.269062
0.268964
0.268871
0.268783
0.268699
0.26862
0.268544
0.268472
0.268404
0.268339
0.268277
0.268218
0.268161
0.268107
0.268056
0.268007
0.26796
0.267916
0.267873
0.267832
0.267794
0.267756
0.267721
0.267687
0.267654
0.267623
0.267594
0.267565
0.267538
0.267512
0.267487
0.267464
0.267441
0.267419
0.267398
0.267378
0.267359
0.267341
0.267324
0.267307
0.267291
0.267276
0.267261
0.267247
0.267234
0.267221
0.267209
0.267197
0.267186
0.267176
0.267165
0.267156
0.267146
0.267137
0.267129
0.267121
0.267113
0.267105
0.267098
0.267092
0.267085
0.267079
0.267073
0.267067
0.267062
0.267057
0.267052
0.267047
0.267043
0.267039
0.267034
0.267031
0.267027
0.267023
0.26702
0.267017
0.267014
0.267011
0.267008
0.267006
0.267003
0.267001
0.266998
0.266996
0.266994
0.266992
0.26699
0.266989
0.266987
0.266985
0.266984
0.266982
0.266981
0.26698
0.266979
0.266978
0.266976
0.266975
0.266974



<matplotlib.figure.Figure at 0x7f8c0037e890>