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

#*****************************************************************************80
#
## faithful_kmeans does a simple clustering exercise.
#
#  Discussion:
#
#    Clustering data.
#
#  Licensing:
#
#    This code is distributed under the GNU LGPL license.
#
#  Modified:
#
#    19 September 2019
#
#  Author:
#
#    John Burkardt
#
  import matplotlib.pyplot as plt
  import numpy as np
  import platform

  print ( '' )
  print ( 'faithful_kmeans:' )
  print ( '  Python version: %s' % ( platform.python_version ( ) ) )
  print ( '  MATH1900 Selected Topics in Mathematics: Machine Learning' )
#
#  Read the data.
#
  data = np.loadtxt ( 'faithful_data.txt' )
#
#  Create x and y.
#
  x = data[:,0]
  y = data[:,1]
  n = len ( x )
  print ( '' )
  print ( '  Number of data values is %d' % ( n ) )
#
#  Normalize the data.
#
  xmin = np.min ( x )
  xmax = np.max ( x )
  ymin = np.min ( y )
  ymax = np.max ( y )
  data[:,0] = ( data[:,0] - xmin ) / ( xmax - xmin )
  data[:,1] = ( data[:,1] - ymin ) / ( ymax - ymin )

  x = data[:,0]
  y = data[:,1]

  plt.plot ( x, y, 'k.', markersize = 10 )
  plt.xlabel ( '<-- Duration (normalized) -->', fontsize = 16 )
  plt.ylabel ( '<-- Wait (normalized) -->', fontsize = 16 )
  plt.title ( 'Old Faithful eruption durations and waits', fontsize = 16 )
  plt.grid ( True )
  plt.axis ( 'equal' )
  filename = 'faithful_data.png'
  plt.savefig ( filename )
  plt.show ( )
  plt.clf ( )
  print ( '' )
  print ( '  Graphics saved as "%s"' % ( filename ) )
#
#  Determine clusters using arbitrary centers.
#
  c = np.array ( [
    [ 0.4, 0.6 ],
    [ 0.6, 0.3 ] ] )

  print ( '  Initial cluster centers C:' )
  print ( c )

  bd = np.sqrt ( ( x - c[0,0] )**2 + ( y - c[0,1] )**2 )
  rd = np.sqrt ( ( x - c[1,0] )**2 + ( y - c[1,1] )**2 )
  bc = np.where ( bd < rd )
  rc = np.where ( rd < bd )
#
#  Compute and print cost.
#
  cost = sum ( bd[bc] ) + sum ( rd[rc] )
  print ( '  Cost of initial clustering is ', cost )

  plt.plot ( x[bc], y[bc], 'c.', markersize = 10 )
  plt.plot ( x[rc], y[rc], 'm.', markersize = 10 )
  plt.plot ( c[0,0], c[0,1], 'bo', markersize = 15 )
  plt.plot ( c[1,0], c[1,1], 'ro', markersize = 15 )
  plt.xlabel ( '<-- Duration -->', fontsize = 16 )
  plt.ylabel ( '<-- Wait -->', fontsize = 16 )
  plt.title ( 'Clusters after 1 step of kmeans', fontsize = 16 )
  plt.grid ( True )
  filename = 'faithful_bluered.png'
  plt.savefig ( filename )
  plt.show ( )
  plt.clf ( )
  print ( '' )
  print ( '  Graphics saved as "%s"' % ( filename ) )
#
#  Prepare for next step by updating cluster centers.
#
  c[0,0] = np.mean ( x[bc] )
  c[0,1] = np.mean ( y[bc] )
  c[1,0] = np.mean ( x[rc] )
  c[1,1] = np.mean ( y[rc] )

  print ( '  Next cluster centers C would be:' )
  print ( c )

  print ( '' )
  print ( '  We will terminate this algorithm after just one step.' )
#
#  Terminate.
#
  print ( '' )
  print ( 'faithful_kmeans:' )
  print ( '  Normal end of execution.' )

  return

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