Matlab KMean Clustering Example

kmeans.m

%{
  Matlab Kmean example
  @author Loreto Parisi (loretoparisi at musixmatch dot com)
  @2015-2016 Musixmatch Spa.
%}

%{ clean all figures and dock new figure }
close all;
figure
try 
    set(figure(1), 'DockControls', 'on');
    jFrame = get(handle(figure(1)), 'JavaFrame');
    jFrame.fHG1Client.setClientWindowStyle(1,0);
catch
end

XDATA=[randn(100,2)*0.75+ones(100,2),randn(100,2)*0.5+ones(100,2)];

%{ 4 clusters kmeans }
[idx,C] = kmeans(NN,4,'Distance','cityblock','Replicates',5,'Options',statset('Display','final'));

plot(XDATA(idx==1,1),NN(idx==1,2),'red.','MarkerSize',12);
hold on;
plot(XDATA(idx==2,1),NN(idx==2,2),'green.','MarkerSize',12);
plot(XDATA(idx==3,1),NN(idx==3,2),'blue.','MarkerSize',12);
plot(XDATA(idx==4,1),NN(idx==4,2),'yellow.','MarkerSize',12);

%{ centroids }
plot(C(:,1),C(:,2),'kx','MarkerSize',9,'LineWidth',3);


%{ legend }
legend('Cluster 1','Cluster 2','Cluster 3','Cluster 4','Centroids','Location','NW')

The script will generate a 4 dimensions vector of random numbers to calculate Kmeans clustering:

[idx,C] = kmeans(NN,4,'Distance','cityblock','Replicates',5,'Options',statset('Display','final'));

Centroid plotting is done as well:

plot(C(:,1),C(:,2),'kx','MarkerSize',9,'LineWidth',3);