#! /usr/bin/env python3
#
from sklearn import *

def study ( ):

#*****************************************************************************80
#
## study is an artificial logistic regression example using six points.
#
#  Licensing:
#
#    This code is distributed under the MIT license.
#
#  Modified:
#
#    27 January 2020
#
#  Author:
#
#    John Burkardt
#
  import numpy as np
  import matplotlib.pyplot as plt
  import platform
  import sklearn
  from sklearn import linear_model

  print ( '' )
  print ( 'study' )
  print ( '  python version: %s' % ( platform.python_version ( ) ) )
  print ( '  scikit-learn version is %s' % ( sklearn.__version__ ) )
  print ( '  Use sklearn LogisticRegression() on the study-time example.' )
  print ( '' )

  x = np.array( [ 
   0.50, 0.75, 1.00, 1.25, 1.50, 
   1.75, 1.75, 2.00, 2.25, 2.50, 
   2.75, 3.00, 3.25, 3.50, 4.00, 
   4.25, 4.50, 4.75, 5.00, 5.00 ] )
  xx = np.reshape ( x, [20,1] )
  y = np.array ( [ 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1, 1, 1 ] )
# xx = np.reshape ( x, [6,1] )
  lr = linear_model.LogisticRegression ( )
  lr.fit ( xx, y )
  lr_coef = np.ndarray.flatten ( lr.coef_ )
  lr_intercept = np.ndarray.flatten ( lr.intercept_ )
  print ( lr_coef, lr_intercept )
  cutoff = - lr_intercept / lr_coef
  print ( "cutoff", cutoff )
  print ( "score", lr.score(xx,y) )
  print ( "mean", np.mean(y) )
  plt.scatter ( x, y )
  x_eval = np.linspace ( 0, 6, 100 )
  y_eval = logistic_fun ( x_eval * lr_coef + lr_intercept )
  plt.plot ( x_eval, y_eval )
  plt.axvline ( x = cutoff, linestyle = '--', color = 'red' )
  plt.title ( 'study logistic regression example' )
  plt.grid ( True )
  filename = 'study.png'
  plt.savefig ( filename )
  print ( '' )
  print ( '  Graphics saved as "%s"' % ( filename ) )
  plt.show ( )

  print ( '' )
  print ( 'study' )
  print ( '  Normal end of execution.' )

  return

def logistic_fun ( x ):

#*****************************************************************************80
#
## logistic_fun evaluates the logistic function.
#
#  Licensing:
#
#    This code is distributed under the MIT license.
#
#  Modified:
#
#    08 January 2020
#
#  Author:
#
#    John Burkardt
#
#  Input:
#
#    real x, the arguments.
#
#  Output:
#
#    real value, the logistic function evaluated at x.
#
  import numpy as np

  value = 1.0 / ( 1.0 + np.exp ( - x ) );

  return value

if ( __name__ == '__main__' ):
  study ( )