If you search with the Vignettes button checked at the RSiteSearch
destination for "prediction error sums of squares" you find that the
mixOmics package provides a function that returns PRESS. Whether that
is a recommended approach to model scoring seems questionable,
however. It looks very much like a jack-knife or LOOCV and those are
known to be sub-optimal. The authors of the vignette do not recommend
it, either.
I do not know what P2 might represent. IANAS.
--
David.
On May 25, 2010, at 10:30 AM, Alex Roy wrote:
> Hello all,
> Is there any function in R by which I can calculate
> PRESS and
> P2 statistics for linear regression in R?
>
> Thanks
>
> Alex
David Winsemius, MD
West Hartford, CT