[R] multidimensional splines

[Bert Gunter]

1. I do not think what the OP requested exists, at least as I
understand her  -- how would one define the "surfaces" where the
piecewise multidimensional polynomials are "joined"? (only 1-d space
is ordered)

2. But, as she has already been told, there are various ways of doing
multivariate smoothing splines. The mars() function in packages mda
and earth is another. Apparently the one in earth adds "extra
features" to the one in mda.

Cheers,
Bert


Bert Gunter

"The trouble with having an open mind is that people keep coming along
and sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )


On Fri, Dec 18, 2015 at 10:52 AM, David Winsemius
<dwinsemius at comcast.net> wrote:
>
>> On Dec 18, 2015, at 7:42 AM, <francesca.mancini at ubs.com> <francesca.mancini at ubs.com> wrote:
>>
>> Dear all,
>>
>> I am looking for a package which performs splining in more than one dimension with polynomials of order>1, exactly as the package "splines" does in 1D.
>> For the moment I was only able to find the package "polspline", whose function "polymars" performs just piecewise linear splines, so not exactly what I'm looking for.
>> Do you know of anything else?
>
> I have pleasing experience with the `rcs` function from package 'rms'. One can build such multidimensional splines with the formula interface using the "*" crossing-operator. It is limited to only piecewise cubic splines. You can specify the knots or let them be optimized by the regression function. It's effective use does require that you learn the requirements of the surrounding helper functions and data description methods supplied by the rms/Hmisc packages.
>
>
> --
> David
>
>>
>
> David Winsemius
> Alameda, CA, USA
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> 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.