AFAIK they are not available, both in principle (the packages fit at a
smaller set of points and interpolate) and in practice (the fits at the
points where fitting is used are not retained in the Fortran/C code).
So it looks like the designers saw no need for these to be available, and
neither do I. Given that loess and locfit are rather different (and
KernSmooth has locpoly too), *why* do you want these components of the
fits?
On Wed, 9 Apr 2003, Deli Wang wrote:
> Could someone tell me is there any way to get the local weight , local
> parameters estimation for each points when using LOESS or locfit packages?
--
Brian D. Ripley, ripley at stats.ox.ac.uk
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UK Fax: +44 1865 272595