brede_mat_cca

(export/brede/brede_mat_cca.m)

Function Synopsis

[A,S,B] = brede_mat_cca(X, Y, varargin)

Help text

 brede_mat_cca        - Canonical correlation analysis

       [A,S,B] = brede_mat_cca(X, Y, 'PropertyName', 'PropertyValue') 
 
       Input:    X   'Mat' structure
                 Y   'Mat' structure
 
       Property: Centering       [ {None} | Both ]
                 InputComponents [ All | {EigenvaluesAboveMean} | a
                                 positive integer ]  
                 Nuisance        'Mat' structure with
                                 nuisance/covariates 

       Output:   A   Left canonical correlation vectors
                 S   Canonical correlation coefficients
                 B   Right canonical correlation vectors

       Canonical correlation analysis between two matrices. The two
       input matrices should have the same number rows.

       By default the data is not centered (centering=none).
 
       By default the data is projected to a subspace by singular
       value decomposision ('InputComponents'='EigenvaluesAboveMean')
       and only the subspace associated with principal component
       eigenvalues above or equal to their mean are included in the
       further canonical correlation analysis. 

       If the 'nuisance' property is defined a nuisance subspace will
       be removed from X and Y for covariance-adjusted canonical
       correlation analysis to obtain partial canonical correlation
       coefficients.

       Example:
         % Example from Mardia 1979
         load(fullfile(fileparts(which('brede')), 'data', 'FretsG1921Heredity.txt'))
         X = FretsG1921Heredity(:,1:2);
         Y = FretsG1921Heredity(:,3:4);
         [A,S,B] = brede_mat_cca(X,Y, 'centering', 'both', 'inputcomponents', 'all');
         diag(S.matrix)  % These should be approx. 0.7886 and 0.0539.

         % Text mining example
         Mvoxel = brede_bib_bib2mat(B, 'type', 'Exp.Loc.coord');
         Mterm0 = brede_bib_bib2mat(B, 'type', 'abstract');
         M = brede_mat_elimsingle(Mterm0);
         M = brede_mat_elimstop(M, 'filename', 'stop_english1.txt');
         M = brede_mat_elimstop(M, 'filename', 'stop_medline.txt');
         M = brede_mat_elimstop(M, 'filename', 'stop_meshcommon.txt');
         M = brede_mat_elimstop(M, 'filename', 'stop_pubmed_neg1.txt');
         Mterm = brede_mat_scale(M, 'type', 'idf');
         [A,S,C] = brede_mat_cca(Mvoxel, Mterm)
         brede_mat_2mat2html(A, C, 'filename', 'cca.html');

       See also BREDE, BREDE_MAT, BREDE_MAT_IBFA, BREDE_MAT_KMC,
                BREDE_MAT_ORTHOSPACE, BREDE_MAT_SVD, BREDE_MAT_NMF.  

 $Id: brede_mat_cca.m,v 1.10 2007/11/08 22:54:35 fn Exp $

Cross-Reference Information

brede_mat_centerexport/brede/brede_mat_center.m
brede_mat_mat2matexport/brede/brede_mat_mat2mat.m
brede_mat_orthospaceexport/brede/brede_mat_orthospace.m
brede_mat_svd  export/brede/brede_mat_svd.m