%
% Examples 9.15 and 9.17
% 
% epage 392 (cluster/
% 
% Written by:
% -- 
% John L. Weatherwax                2008-02-20
% 
% email: wax@alum.mit.edu
% 
% Please send comments and especially bug reports to the
% above email address.
% 
%-----

clear all; 
close all; 
clc; 

addpath('../../Code/CSTool');
load measure; 

% lump everything into one data matrix: 
X = measure; 
X = zscore(X);

%--
% Apply Hierarchical clustering (epage 382): 
%--
if( 1 ) 
  ye = pdist(X,'euclid'); 
  %ye = pdist(X,'cityblock');
  z = linkage(ye,'single');
  %z = linkage(ye,'complete'); 
  %z = linkage(ye,'average'); 
  %z = linkage(ye,'ward'); 
  [H,T] = dendrogram(z); 
  xlabel('datum instance'); 
  title( 'algomative clustering with single linkage' );
  saveas( gcf, 'prob_9_4_hierarchical_linkage', 'epsc' ); 

  fprintf('the cophenet coefficient for this clustering is = %10.5f\n',cophenet(z,ye)); 
end

%--
% Apply k-means clustering ()
%--
if( 1 ) 
  k = 2;           % <- assume we have two clusters 
  [cid,nr,centers] = cskmeans(X,k); 

  % plot the two clusters overlayed with truth: 
  figure; 
  tm=plot( X(1:10,1), X(1:10,2), 'sk' ); hold on;    % <- this is the truth (assume the first 10 data points are men)
  tw=plot( X(11:20,1), X(11:20,2), 'dk' ); 

  men_cluster = cid(7);                              % <- this is what MATLAB called the mens cluster ... 
  if( men_cluster==1 )                               % <- so this must be what MATLAB called the womans cluster ... 
    women_cluster=2; 
  else
    women_cluster=1; 
  end

  am=plot( X(find(cid==men_cluster),1), X(find(cid==men_cluster),2), 'sr', 'markersize', 10 ); 
  aw=plot( X(find(cid==women_cluster),1), X(find(cid==women_cluster),2), 'dr', 'markersize', 10 ); 
  legend( [tm, tw, am, aw], { 'true men', 'true women', 'approx men', 'approx women' }, 'location', 'best' ); 
  saveas( gcf, 'prob_9_4_k_means', 'epsc' ); 
end