gam() algorithm

["Kunio Takezawa" <s.otasuke@gmail.com>]

S-Plus Users
E-mail: s-news@lists.biostat.wustl.edu

   Hi!
   My question is on gam().
   I have thought that gam() uses backfitting to derive an additive model.
But after I browsed the gam() routine, I have some doubts about it, though
I have not understand the routine totally. My guess is that it solves multiple
linear regression using QR decomposition. Iterations are carried out for
iterative reweighted least squares not for deriving coefficients of an additive
model.
   If my guess is not correct, please let me know where the backfitting
is carried out in gam().