featsel.m

这个为模式识别工具箱

%FEATSEL Selection of known features
%
%   W = FEATSEL(K,J)
%
% INPUT
%   K    Input dimensionality
%   J    Index vector of features to be selected
%
% OUTPUT
%   W    Mapping performing the feature selection
%
% DESCRIPTION
% This is a simple support routine that writes feature selection
% in terms of a mapping. If A is a K-dimensional dataset and J are
% the feature indices to be selected, then B = A*W does the same as
% B = A(I,J).
%
% The use of this routine is a mapping V computed for a lower dimensional
% subspace defined by J can now be defined by W = FEATSEL(K,J)*V as a 
% mapping in the original K-dimensional space.
%
% The selected features can be retrieved by W.DATA or by +W.
% See below for various methods to perform feature selection.
%
% SEE ALSO
% MAPPINGS, DATASETS, FEATEVAL, FEATSELF, FEATSELLR,
% FEATSELO, FEATSELB, FEATSELI, FEATSELP, FEATSELM, PRPROGRESS

% Copyright: R.P.W. Duin, [e-mail]
% Faculty EWI, Delft University of Technology
% P.O. Box 5031, 2600 GA Delft, The Netherlands

function w = featsel(k,j)

if isdataset(k) & ismapping(j)
	J = j.data;
	w = k(:,J);
else
	if (any(j) > k | any(j) < 1)
		error('Features to be selected are not in proper range')
	end
	w = mapping(mfilename,'trained',j(:)',[],k,length(j));
	w = setname(w,'Feature Selection');
end