Daniel@0: %DEMKNN1 Demonstrate nearest neighbour classifier.
Daniel@0: %
Daniel@0: %	Description
Daniel@0: %	The problem consists of data in a two-dimensional space.  The data is
Daniel@0: %	drawn from three spherical Gaussian distributions with priors 0.3,
Daniel@0: %	0.5 and 0.2; centres (2, 3.5), (0, 0) and (0,2); and standard
Daniel@0: %	deviations 0.2, 0.5 and 1.0. The first figure contains a scatter plot
Daniel@0: %	of the data.  The data is the same as in DEMGMM1.
Daniel@0: %
Daniel@0: %	The second figure shows the data labelled with the corresponding
Daniel@0: %	class given by the classifier.
Daniel@0: %
Daniel@0: %	See also
Daniel@0: %	DEM2DDAT, DEMGMM1, KNN
Daniel@0: %
Daniel@0: 
Daniel@0: %	Copyright (c) Ian T Nabney (1996-2001)
Daniel@0: 
Daniel@0: clc
Daniel@0: disp('This program demonstrates the use of the K nearest neighbour algorithm.')
Daniel@0: disp(' ')
Daniel@0: disp('Press any key to continue.')
Daniel@0: pause
Daniel@0: % Generate the test data
Daniel@0: ndata = 250;
Daniel@0: randn('state', 42);
Daniel@0: rand('state', 42);
Daniel@0: 
Daniel@0: [data, c] = dem2ddat(ndata);
Daniel@0: 
Daniel@0: % Randomise data order
Daniel@0: data = data(randperm(ndata),:);
Daniel@0: 
Daniel@0: clc
Daniel@0: disp('We generate the data in two-dimensional space from a mixture of')
Daniel@0: disp('three spherical Gaussians. The centres are shown as black crosses')
Daniel@0: disp('in the plot.')
Daniel@0: disp(' ')
Daniel@0: disp('Press any key to continue.')
Daniel@0: pause
Daniel@0: fh1 = figure;
Daniel@0: plot(data(:, 1), data(:, 2), 'o')
Daniel@0: set(gca, 'Box', 'on')
Daniel@0: hold on
Daniel@0: title('Data')
Daniel@0: hp1 = plot(c(:, 1), c(:,2), 'k+')
Daniel@0: % Increase size of crosses
Daniel@0: set(hp1, 'MarkerSize', 8);
Daniel@0: set(hp1, 'LineWidth', 2);
Daniel@0: hold off
Daniel@0: 
Daniel@0: clc
Daniel@0: disp('We next use the centres as training examplars for the K nearest')
Daniel@0: disp('neighbour algorithm.')
Daniel@0: disp(' ')
Daniel@0: disp('Press any key to continue.')
Daniel@0: pause
Daniel@0: 
Daniel@0: % Use centres as training data
Daniel@0: train_labels = [1, 0, 0; 0, 1, 0; 0, 0, 1];
Daniel@0: 
Daniel@0: % Label the test data up to kmax neighbours
Daniel@0: kmax = 1;
Daniel@0: net = knn(2, 3, kmax, c, train_labels);
Daniel@0: [y, l] = knnfwd(net, data);
Daniel@0: 
Daniel@0: clc
Daniel@0: disp('We now plot each data point coloured according to its classification.')
Daniel@0: disp(' ')
Daniel@0: disp('Press any key to continue.')
Daniel@0: pause
Daniel@0: % Plot the result
Daniel@0: fh2 = figure;
Daniel@0: colors = ['b.'; 'r.'; 'g.'];
Daniel@0: for i = 1:3
Daniel@0:   thisX = data(l == i,1);
Daniel@0:   thisY = data(l == i,2);
Daniel@0:   hp(i) = plot(thisX, thisY, colors(i,:));
Daniel@0:   set(hp(i), 'MarkerSize', 12);
Daniel@0:   if i == 1
Daniel@0:     hold on
Daniel@0:   end
Daniel@0: end
Daniel@0: set(gca, 'Box', 'on');
Daniel@0: legend('Class 1', 'Class 2', 'Class 3', 2)
Daniel@0: hold on
Daniel@0: labels = ['1', '2', '3'];
Daniel@0: hp2 = plot(c(:, 1), c(:,2), 'k+');
Daniel@0: % Increase size of crosses
Daniel@0: set(hp2, 'MarkerSize', 8);
Daniel@0: set(hp2, 'LineWidth', 2);
Daniel@0: 
Daniel@0: test_labels = labels(l(:,1));
Daniel@0: 
Daniel@0: title('Training data and data labels')
Daniel@0: hold off
Daniel@0: 
Daniel@0: disp('The demonstration is now complete: press any key to exit.')
Daniel@0: pause
Daniel@0: close(fh1);
Daniel@0: close(fh2);
Daniel@0: clear all; 
Daniel@0: