featseli.m

The pattern recognition matlab toolbox

%FEATSELI Individual feature selection for classification
% 
% [W,R] = FEATSELI(A,CRIT,K,T)
% 
% INPUT	
%   A    Training dataset
%   CRIT Name of the criterion or untrained mapping
%        (default: 'NN', i.e. the 1-Nearest Neighbor error)
%   K    Number of features to select (default: sort all features)
%   T    Tuning dataset (optional)
%
% OUTPUT
%   W    Feature selection mapping
%   R    Matrix with criterion values
%
% DESCRIPTION
% Individual selection of K features using the dataset A. CRIT sets the
% criterion used by the feature evaluation routine FEATEVAL. If the dataset
% T is given, it is used as test set for FEATEVAL. For K = 0 all features are
% selected, but reordered according to the criterion. The result W can be
% used for selecting features using B*W. 
% The selected features are stored in W.DATA and can be found by +W.
% In R, the search is reported step by step as:
% 
% 	R(:,1) : number of features
% 	R(:,2) : criterion value
% 	R(:,3) : added / deleted feature
% 
% SEE ALSO
% MAPPINGS, DATASETS, FEATEVAL, FEATSELO, FEATSELB, FEATSELF,
% FEATSEL,  FEATSELP, FEATSELM

% Copyright: R.P.W. Duin, [e-mail]
% Faculty of Applied Sciences, Delft University of Technology
% P.O. Box 5046, 2600 GA Delft, The Netherlands

% $Id: featseli.m,v 1.3 2007/04/21 23:06:46 duin Exp $

function [w,r] = featseli(a,crit,ksel,t)

	prtrace(mfilename);

	if (nargin < 2) | isempty(crit)
    prwarning(2,'no criterion specified, assuming NN');
		crit = 'NN'; 
	end
	if (nargin < 3 | isempty(ksel))
		ksel = 0; 
	end
	if (nargin < 4)
    prwarning(3,'no tuning set supplied (risk of overfit)');
		t = [];     
	end

  % If no arguments are supplied, return an untrained mapping.

	if (nargin == 0) | (isempty(a))
		w = mapping('featseli',{crit,ksel,t});
		w = setname(w,'Individual FeatSel');
		return
	end

	[m,k,c] = getsize(a); featlist = getfeatlab(a);

	% If KSEL is not given, return all features.

	if (ksel == 0), ksel = k; end
	
	isvaldfile(a,1,2); % at least 1 object per class, 2 classes
	a = testdatasize(a);
	iscomdset(a,t);
	
	critval = zeros(k,1);
	
	% Evaluate each feature in turn.

	if (isempty(t))
		for j = 1:k
			critval(j) = feateval(a(:,j),crit);
		end
	else
		for j = 1:k
			critval(j) = feateval(a(:,j),crit,t(:,j));
		end
	end

	% Sort the features by criterion value (maximum first).
	[critval_sorted,J] = sort(-critval);
	r = [[1:k]', -critval_sorted, J];
	J = J(1:ksel)'; 

	% Return the mapping found.
	w = featsel(k,J);
	w = setlabels(w,featlist(J,:));
	w = setname(w,'Individual FeatSel');

return