#! /usr/bin/env python3
#
def moon_classify_mlp ( ):

#*****************************************************************************80
#
## moon_classify_mlp() uses a multilayer perceptron to classify the moon data.
#
#  Licensing:
#
#    This code is distributed under the MIT license.
#
#  Modified:
#
#    05 August 2023
#
#  Author:
#
#    Andreas Mueller, Sarah Guido.
#    This version by John Burkardt.
#
#  Reference:
#
#    Andreas Mueller, Sarah Guido,
#    Introduction to Machine Learning with Python,
#    OReilly, 2017,
#    ISBN: 978-1-449-36941-5
#
  from sklearn.datasets import make_moons
  from sklearn.neural_network import MLPClassifier
  from sklearn.model_selection import train_test_split
  import matplotlib.pyplot as plt
  import mglearn
  import numpy as np
  import platform
  import sklearn

  print ( '' )
  print ( 'moon_classify_mlp():' )
  print ( '  Python version: ' + platform.python_version ( ) )
  print ( '  scikit-learn version: '+ sklearn.__version__ )
  print ( '  Generate 100 samples of the artificial moon dataset.' )
  print ( '  Classify the data using a multilayer perceptron.' )
#
#  Create the data.
#
  X, y = make_moons ( 
    n_samples = 100,
    noise = 0.25,
    random_state = 3 )
#
#  Split the data into training and test sets.
#
  X_train, X_test, y_train, y_test = train_test_split ( \
    X, y, stratify = y, random_state = 42 )
#
#  Create an MLP classifier and fit it to the training data.
#  By default, 100 hidden nodes are used.
#
  mlp = MLPClassifier ( 
    solver = 'lbfgs', 
    random_state = 0 )

  mlp.fit ( X_train, y_train )
#
#  Display the data.
#
  plt.clf ( )

  mglearn.plots.plot_2d_separator ( 
    mlp,
    X_train,
    fill = True,
    alpha = 0.3 )

  mglearn.discrete_scatter ( X_train[:,0], X_train[:,1], y_train )
  plt.grid ( True )
  plt.xlabel ( "Feature 0" )
  plt.ylabel ( "Feature 1" )
  plt.title ( "MLP classifies moon data (100 nodes)" )
  filename = 'moon_classify_mlp_100.png'
  plt.savefig ( filename )
  print ( '  Graphics saved as "' + filename + '"' )
#
#  Repeat, but use just 10 hidden nodes.
#
  mlp = MLPClassifier ( 
    solver = 'lbfgs', 
    random_state = 0, 
    hidden_layer_sizes = [10],
    max_iter = 500 )

  mlp.fit ( X_train, y_train )
#
#  Display the data.
#
  plt.clf ( )

  mglearn.plots.plot_2d_separator ( 
    mlp,
    X_train,
    fill = True,
    alpha = 0.3 )

  mglearn.discrete_scatter ( X_train[:,0], X_train[:,1], y_train )
  plt.grid ( True )
  plt.xlabel ( "Feature 0" )
  plt.ylabel ( "Feature 1" )
  plt.title ( "MLP classifies moon data (10 nodes)" )
  filename = 'moon_classify_mlp_010.png'
  plt.savefig ( filename )
  print ( '  Graphics saved as "' + filename + '"' )
#
#  Repeat, but use two layers of 10 nodes.
#
  mlp = MLPClassifier ( 
    solver = 'lbfgs', 
    random_state = 0, 
    hidden_layer_sizes = [10,10],
    max_iter = 500 )

  mlp.fit ( X_train, y_train )
#
#  Display the data.
#
  plt.clf ( )

  mglearn.plots.plot_2d_separator ( 
    mlp,
    X_train,
    fill = True,
    alpha = 0.3 )

  mglearn.discrete_scatter ( X_train[:,0], X_train[:,1], y_train )
  plt.grid ( True )
  plt.xlabel ( "Feature 0" )
  plt.ylabel ( "Feature 1" )
  plt.title ( "MLP classifies moon data (10+10 nodes)" )
  filename = 'moon_classify_mlp_010_010.png'
  plt.savefig ( filename )
  print ( '  Graphics saved as "' + filename + '"' )
#
#  Repeat, but use two layers of 10 nodes and tanh nonlinearity.
#
  mlp = MLPClassifier ( 
    solver = 'lbfgs', 
    random_state = 0, 
    hidden_layer_sizes = [10,10],
    activation = 'tanh',
    max_iter = 500 )

  mlp.fit ( X_train, y_train )
#
#  Display the data.
#
  plt.clf ( )

  mglearn.plots.plot_2d_separator ( 
    mlp,
    X_train,
    fill = True,
    alpha = 0.3 )

  mglearn.discrete_scatter ( X_train[:,0], X_train[:,1], y_train )
  plt.grid ( True )
  plt.xlabel ( "Feature 0" )
  plt.ylabel ( "Feature 1" )
  plt.title ( "MLP classifies moon data (10+10 nodes + tanh)" )
  filename = 'moon_classify_mlp_010_010_tanh.png'
  plt.savefig ( filename )
  print ( '  Graphics saved as "' + filename + '"' )
#
#  Terminate.
#
  print ( '' )
  print ( 'moon_classify_mlp():' )
  print ( '  Normal end of execution.' )

  return

def timestamp ( ):

#*****************************************************************************80
#
## timestamp() prints the date as a timestamp.
#
#  Licensing:
#
#    This code is distributed under the MIT license. 
#
#  Modified:
#
#    21 August 2019
#
#  Author:
#
#    John Burkardt
#
  import time

  t = time.time ( )
  print ( time.ctime ( t ) )

  return

if ( __name__ == '__main__' ):
  timestamp ( )
  moon_classify_mlp ( )
  timestamp ( )