plstrain.m

有监督自组织映射-偏最小二乘算法(A supervised self-organising map–partial least squares algorithm),可以用语多变量数据的回归分析

function [B,XLoadings,MLKP] = PlsTrain(Xtrain, Ytrain, MLKP);
%
% Calculates a model based on Multiple Linear Regression
%
% Format: [B, Xloadings, MLKP] = plstrain(Xtrain, Ytrain, MLKP)
%
% Xtrain: input data of the training set
% Ytrain: output data of the traning set
% MLKP:   parameter structure
%
% B:         matrix containing the regression coefficients based on Xtrain and Ytrain
% XLoadings: the loadings for the X-block (necessary for analysing SOMPLS)
% MLKP:      the parameter structure including in case of SIMPLS the mean
%            values of the input and outpur matrices in the training set
%
warning off
[Nobj,NvarX] = size(Xtrain);
[Nobj,NvarY] = size(Ytrain);
if MLKP.PLSMode == 'SIMPLS'
    MLKP.PLSMX = mean(Xtrain);
    XTrainMC = (Xtrain-MLKP.PLSMX(ones(Nobj,1),:));
    MLKP.PLSMY = mean(Ytrain);
    YTrainMC = (Ytrain-MLKP.PLSMY(ones(Nobj,1),:));
    [B,C,XLoadings,T,U,R,R2X,R2Y]=SIMPLS(Xtrain,Ytrain,MLKP.LatentVar);
else
    [B,XLoadings] = NIPPLS(Xtrain,Ytrain,MLKP.LatentVar);
end