Subscribe Bookmark RSS Feed

Cross validation wants to use 0 factors in PLS


Community Trekker


Apr 15, 2015


I am quite new to JMP/SAS, and I am working with a bunch of variables, trying to make some kind of a model. So I heard that a PLS is quite a handy way to deal with large amounts of data. For some subsets I get that the minimum number of factors is 8 (or another value) (NIPALS algorithm), and from those I can get VIP vs Coefficients plots etc and see which factors are important. But sometimes the cross validation give a van der Voet T2 value/P-value of 0,0000/1,0000, and it says (e.g.) that "Note. The minimum root mean PRESS is 1,07143 and the minimizing numbers of factors is 0." From those I obviously can't get any more information.

So, my question is, why do I get those minimum number of factors to be 0. What does it "mean", and is there something to do about it or is it just to accept that that subset doesn't have any important information?

Cheers, Sebastian