s-news
[Top] [All Lists]

smoothing matrix

To: s-news@wubios.wustl.edu
Subject: smoothing matrix
From: "Dr. L. Y. Hin" <lyhin@netvigator.com>
Date: Fri, 21 Dec 2001 00:14:37 +0800
References: <3.0.3.32.20011031164730.007caaa0@pop.netvigator.com> <001b01c16227$77acbb60$d6bb42ab@GIBBERS> <3BF10937.F395D62D@netvigator.com> <002401c16cca$b56c6ce0$d6bb42ab@GIBBERS>
Dear S-plus users,


I have a problem with smoothing spline that I'd be grateful for your
advice.
I've been advised that I can obtain the smoother matrix, ie., S=(I +
lambda*K)^-1, from the smooth.spline() function as below:

x_1:20
hat_diag(x)
spar_smooth.spline(x)$spar
for (i in 1:length(x))
hat[,i]_smooth.spline(smooth.spline,hat[,i],spar=spar)$y

and hat should be the smoother matrix S.


However, this does not work for cases with ties, eg.,

x_rep(1:5,(2:6))

Kindly please advise me whether there is any way for me to obtain the
smoother matrix S as shown in the Reinsch form above?

Thank you. 

Best
Lin

<Prev in Thread] Current Thread [Next in Thread>