pls_transform.m

The pattern recognition matlab toolbox

%pls_transform  Partial Least Squares transformation
%
%  T = pls_transform(X,R)
%  T = pls_transform(X,R,Options)
%
% INPUT
%  X       [N -by- d_X]    the input  data matrix, N samples, d_X variables
%  R       [d_X -by- nLV]  the transformation matrix: T_new = X_new*R
%          (X_new here after preprocessing, preprocessing and un-preprocessing 
%          could be done automatically (than Options contains info about
%          preprocessing) or manually); normally, R as a field of XRes
%          output parameter of pls_train routine
%
%  Options structure returned by pls_train (if not supplied then will be
%  no preprocessing performed)  
%
% OUTPUT
%  T [N -by- nLV]   scores -- transformed data
%
% DESCRIPTION
% Applys PLS (Partial Least Squares) regression model
%
% SEE ALSO
% pls_train, pls_apply

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

% $Id: pls_transform.m,v 1.1 2007/08/28 11:00:39 davidt Exp $

function T = pls_transform(X,R,Options)

if nargin < 3
  Options  = [];
end

DefaultOptions.X_centering = [];
DefaultOptions.Y_centering = [];
DefaultOptions.X_scaling = [];
DefaultOptions.Y_scaling = [];

Options = pls_updstruct(DefaultOptions, Options);

[N, d_X]    = size(X);
[d_XR, nLv] = size(R);

if d_X ~= d_XR
  error('size(X,2) must be equal to size(R,1)');
end

T = pls_prepro(X, Options.X_centering, Options.X_scaling)*R;

return;