I have noticed differences between PLRS matlab vs Eigenvector Research software based on predicted values. For matlab, I am using the below steps; and calculated predicted values by using leave out cross validation method.
C=cvpartition(50,'leaveout')
[Xl,Yl,Xs,Ys,beta,pctVar,mse,stats] = plsregress(X,y,8,'CV',C);
yfit = [ones(size(X,1),1) X]*beta;
On the other hand, I have Eigenvector toolbox and using specific function related Eigenvector Research software to calculated predicted values, see below
[press,cumpress,rmsecv,rmsec,cvpred,misclassed] = crossval(X,y,'sim','loo',8);
Even if I have eactly same regression coefficient for both matlabi and eignevector software, the yfit is not equal with cvpred? They are similat behaviour, but I have differences based on scale. I am sure that I am using mean center preprocess. I would like to learn how cvpred is calculating, I did not find any information on how this value is calculated