%GENDATC Generation of two spherical classes with different variances
%
% A = GENDATC(N,K,U,LABTYPE)
%
% INPUT
% N Vector with class sizes (default: [50,50])
% K Dimensionality of the dataset (default: 2)
% U Mean of class 1 (default: 0)
% LABTYPE 'crisp' or 'soft' labels (default: 'crisp')
%
% OUTPUT
% A Dataset
%
% DESCRIPTION
% Generation of a K-dimensional 2-class dataset A of N objects.
% Both classes are spherically Gaussian distributed.
%
% Class 1 has the identity matrix as covariance matrix and
% mean U. If U is a scalar then [U,0,0,..] is used as class mean.
% Class 2 has also the identity matrix as covariance matrix, except
% for a variance of 4 for the first two features. Its mean is 0.
% Class priors are P(1) = P(2) = 0.5.
%
% If N is a vector of sizes, exactly N(I) objects are generated
% for class I, I = 1,2.
%
% The default means result in a class overlap of 0.16.
%
% LABTYPE defines the desired label type: 'crisp' or 'soft'. In the
% latter case true posterior probabilities are set for the labels.
%
% Defaults: N = [50,50], K = 2, U = 0, LABTYPE = 'crisp'.
%
% SEE ALSO (<a href="http://37steps.com/prtools">PRTools Guide</a>)
% DATASETS, PRDATASETS
% Copyright: R.P.W. Duin, duin@ph.tn.tudelft.nl
% Faculty of Applied Sciences, Delft University of Technology
% P.O. Box 5046, 2600 GA Delft, The Netherlands
% $Id: gendatc.m,v 1.3 2009/01/27 13:01:42 duin Exp $
function A = gendatc(N,k,ma,labtype)
if nargin < 1
N = [50 50];
prwarning(4,'size of classes not specified, assuming [50 50]');
end
if nargin < 2
k=2;
prwarning(4,'dimension not specified, assuming 2');
end
if nargin < 3
ma=0;
prwarning(4,'mean not specified, assuming 0');
end
if nargin < 4
labtype = 'crisp';
prwarning(4,'label type not specified, assuming crisp');
end
p = [0.5 0.5];
N = genclass(N,p);
% When a scalar ma is given, the mean vector should be generated:
if (length(ma) == 1) & (k>1),
ma=[ma,zeros(1,k-1)];
end
GA = eye(k);
GB = eye(k); GB(1,1) = 9;
if k > 1, GB(2,2) = 9; end
mb = zeros(1,k);
U = prdataset([ma;mb],[1 2]','prior',p);
A = gendatgauss(N,U,cat(3,GA,GB));
A = set(A,'name','Spherical Set');
% Take care for the different types of labels:
switch labtype
case 'crisp'
;
case 'soft'
W = nbayesc(U,cat(3,GA,GB));
targets = A*W*classc;
A = setlabtype(A,'soft',targets);
otherwise
error(['Label type ' labtype ' not supported'])
end
return