Vassily Shvets
2012-Mar-12 23:18 UTC
[R] Fwd: Re[2]: B-spline/smooth.basis derivative matrices
--- On Mon, 3/12/12, aleksandr shfets <a_shfets at mail.ru> wrote:> From: aleksandr shfets <a_shfets at mail.ru> > Subject: Fwd: Re[2]: [R] B-spline/smooth.basis derivative matrices > To: "Vassily Shvets" <shv736 at yahoo.com> > Received: Monday, March 12, 2012, 5:15 PM > > > > -------- ???????????? ????????? > -------- > ?? ????: aleksandr shfets <a_shfets at mail.ru> > ????: vito.muggeo <vito.muggeo at unipa.it> > ????: 13 ????? 2012, 03:09 > ????: Re[2]: [R] B-spline/smooth.basis derivative > matrices > > > Hello again,(my message was truncated the first time!) > > Thank you, vito for these functions: I've been able to look > at the smoothing spline matrix and its derivatives that I > constucted with functions "create.bspline.basis" and > "smooth.basis" > > Understanding from Hastie and Tishbirani that the > S(lambda)-- the penalty matrix > > decreases in magnitude compared to the matrix [y[i,j] - > f(x)] would it not be of fundamental importance to > > have the penalty matrix be the same magnitude? > > That is to get both matrices in the smooth, i.e, > > SS(f, ?) =[y ? f(x)]^2 + ?*int(x1,x2)[[f''(x)]^2 dx > > to the same size? > > Especially if one is working with an? R > program(CollocInfer) that will only evaluate yhat =X*beta > with appropriate matrices? > > regards, > s > I hope I don't ask a question whose answer is evident;I > haven't been able to see an answer to this question-- how to > coerce the matrices to be the same size-- from the data > sources I've been able to access. > > 25 ??????? 2012, 15:11 ?? "vito.muggeo" <vito.muggeo at unipa.it>: > > dear s, > > > > If you are willing to use B-splines placing p knots, > say, (rather than smoothing splines > > placing n knots), you could have a look to the > functions splineDesign() or spline.des() > > (argument derivs) in the package "splines". > > > > vito > > > > On Fri, 24 Feb 2012 12:28:58 -0800 (PST), Vassily > Shvets wrote > > > Hello, > > > I've noticed that SPLUS seems to have a function > for evaluating derivative > > > matrices of splines. I've found the R function > that evaluates matrices from > > > 'smooth.spline'; maybe someone has written > something to do the same with smooth.basis? > > > regards, > > > s > > > > > > ______________________________________________ > > > R-help at r-project.org > mailing list > > > https://stat.ethz.ch/mailman/listinfo/r-help > > > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > > > and provide commented, minimal, self-contained, > reproducible code. > > > > -- > > Open WebMail Project (http://openwebmail.org) > > > > > ----------------
Vassily Shvets
2012-Mar-12 23:19 UTC
[R] Fwd: Re[2]: B-spline/smooth.basis derivative matrices
--- On Mon, 3/12/12, aleksandr shfets <a_shfets at mail.ru> wrote:> From: aleksandr shfets <a_shfets at mail.ru> > Subject: Fwd: Re[2]: [R] B-spline/smooth.basis derivative matrices > To: "Vassily Shvets" <shv736 at yahoo.com> > Received: Monday, March 12, 2012, 5:15 PM > > > > -------- ???????????? ????????? > -------- > ?? ????: aleksandr shfets <a_shfets at mail.ru> > ????: vito.muggeo <vito.muggeo at unipa.it> > ????: 13 ????? 2012, 03:09 > ????: Re[2]: [R] B-spline/smooth.basis derivative > matrices > > > Hello again,(my message was truncated the first time!) > > Thank you, vito for these functions: I've been able to look > at the smoothing spline matrix and its derivatives that I > constucted with functions "create.bspline.basis" and > "smooth.basis" > > Understanding from Hastie and Tishbirani that the > S(lambda)-- the penalty matrix > > decreases in magnitude compared to the matrix [y[i,j] - > f(x)] would it not be of fundamental importance to > > have the penalty matrix be the same magnitude? > > That is to get both matrices in the smooth, i.e, > > SS(f, ?) =[y ? f(x)]^2 + ?*int(x1,x2)[[f''(x)]^2 dx > > to the same size? > > Especially if one is working with an? R > program(CollocInfer) that will only evaluate yhat =X*beta > with appropriate matrices? > > regards, > s > I hope I don't ask a question whose answer is evident;I > haven't been able to see an answer to this question-- how to > coerce the matrices to be the same size-- from the data > sources I've been able to access. > > 25 ??????? 2012, 15:11 ?? "vito.muggeo" <vito.muggeo at unipa.it>: > > dear s, > > > > If you are willing to use B-splines placing p knots, > say, (rather than smoothing splines > > placing n knots), you could have a look to the > functions splineDesign() or spline.des() > > (argument derivs) in the package "splines". > > > > vito > > > > On Fri, 24 Feb 2012 12:28:58 -0800 (PST), Vassily > Shvets wrote > > > Hello, > > > I've noticed that SPLUS seems to have a function > for evaluating derivative > > > matrices of splines. I've found the R function > that evaluates matrices from > > > 'smooth.spline'; maybe someone has written > something to do the same with smooth.basis? > > > regards, > > > s > > > > > > ______________________________________________ > > > R-help at r-project.org > mailing list > > > https://stat.ethz.ch/mailman/listinfo/r-help > > > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > > > and provide commented, minimal, self-contained, > reproducible code. > > > > -- > > Open WebMail Project (http://openwebmail.org) > > > > > ----------------