Close, but not quite. Your design matrix is:
H <- bs(X, knots=spl$fit$knot, intercept=TRUE)
So long as you have data in all of the inter-knot regions, you should be ok.
A good book to check out about this sort of thing is Ruppert, Wand and
Carroll's Semiparametric Regression book. I can't recommend that book
highly enough.
Derek
pinkdd wrote:
>
> Thank you. D Sonderegger.
> Did you mean use
>
> bs <- bs(X, knots = spl$fit$knot)
> H <- predict(bs, X)
>
> Then H should be the matrix for the original data under the smoothing
> spline basis?
> However, another problem arises, since I need to use H to estimate the
> coefficient beta, which involves (H'H)^{-1}. But my matrix (H'H) is
> singular, which might be caused by the duplicated data in X. Do you know
> how to fix this problem?
>
> Thanks!
>
>
> D Sonderegger wrote:
>>
>> I believe that smooth.spline fits a cubic B-spline to the data. So you
>> just need to know the knot points (which are returned by smooth.spline as
>> spl$fit$knot) and then use the bs() function in the splines library.
>>
>
>
--
View this message in context:
http://n4.nabble.com/Smoothing-Spline-Basis-Matrix-tp1573131p1574236.html
Sent from the R help mailing list archive at Nabble.com.
______________________________________________
R-help@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.
