[R] PLS Y scores (orthogonal scores algorithm)

L. Shyamal shyamal at camo.co.in
Mon Apr 27 17:28:24 CEST 2009


Can someone make the following change to
http://cran.r-project.org/web/packages/pls/index.html

Line 68: oscorespls.fit.R (2.1-0)
from
            if (nresp == 1)
                break                   # pls1: no iteration

to
           if (nresp == 1)
           {
                u.a <- Y %*% q.a / c(crossprod(q.a))
                break                   # pls1: no iteration
           }
 
so that the Y scores are consistent across algorithms. 

best wishes
Shyamal

Shyamal Lakshminarayanan
Bangalore, India




More information about the R-help mailing list