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

#*****************************************************************************80
#
## apple() classifies fruit as oranges(0) or apples(1).
#
#  Licensing:
#
#    This code is distributed under the GNU LGPL license.
#
#  Modified:
#
#    13 February 2022
#
#  Author:
#
#    John Burkardt
#
  import matplotlib.pyplot as plt
  import numpy as np
  from logistic_regression import logistic_regression

  print ( 'apple():' )
#
#  Read the data file.
#
  data = np.loadtxt ( 'apple_data.txt' )
#
#  First column is weight in grams.  
#  Second column is 0 for orange, 1 for apple.
#
  g = data[:,0]
  y = data[:,1]
  m = len ( g )
#
#  Normalize the G data.
#
  gmin = np.min ( g )
  gmax = np.max ( g )
  gn = ( g - np.min ( g ) ) / ( np.max ( g ) - np.min ( g ) )
#
#  Create a data array X = [ 1 | g ].
#
  X = np.zeros ( [ m, 2 ] )
  X[:,0] = 1
  X[:,1] = gn
#
#  Set the learning rate ALPHA
#
  alpha = 0.02
#
#  Set the number of iterations KMAX.
#
  kmax = 100000
#
#  Compute logistic weights W.
#
  wn = logistic_regression ( X, y, alpha, kmax )

  print ( '' )
  print ( '  Weights for normalized data Wn = (%g,%g)' % ( wn[0], wn[1] ) )

  w = np.array ( [ wn[0] - wn[1] * gmin / ( gmax - gmin ), wn[1] / ( gmax - gmin ) ] )
  print ( '' )
  print ( '  Weights for original data W = (%g,%g)' % ( w[0], w[1] ) )
  cutoff = - w[0] / w[1]
  print ( '  Cutoff value  is ', cutoff )
#
#  Display the data.
#
  gplot = np.linspace ( gmin, gmax, 101 )
  yplot = 1.0 / ( 1.0 + np.exp ( - ( w[0] + w[1] * gplot ) ) )

  plt.plot ( g[y==0], y[y==0], 'mo' )
  plt.plot ( g[y==1], y[y==1], 'go' )
  plt.plot ( gplot, yplot, linewidth = 3 )
  plt.plot ( [cutoff,cutoff], [0,1], '--', color = 'r' )
  plt.grid ( True )
  plt.xlabel ( '<-- G: Weight in grams -->' )
  plt.ylabel ( '<-- Y(G): Orange = 0, Apple = 1 -->' )
  plt.title ( 'Orange/Apple classified by weight' )
  filename = 'apple.png'
  print ( '  Graphics saved as "%s"' % ( filename ) )
  plt.savefig ( filename )
  plt.show ( )
  plt.close ( )
#
#  Terminate.
#
  print ( '' )
  print ( 'apple():' )
  print ( '  Normal end of execution.' )

  return

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