[R] Fit CMARS with R possible (Packages) ?

[varin sacha]

Dear Bert, Dear Marc,

Many thanks for your feedbacks.

Best,






Le samedi 5 janvier 2019 à 00:26:47 UTC+1, Marc Schwartz <marc_schwartz using me.com> a écrit : 





Hi,

Like Bert, I was not able to find anything built in R.

It is possible that CMARS has not yet been implemented in R, or may be in development but not yet ready for release.

I found several references to the use of MOSEK (https://www.mosek.com) along with MATLAB, but both are commercial products.

The MOSEK developers appear to have an Rmosek interface package:

  https://docs.mosek.com/8.1/rmosek/index.html

which provides access to the MOSEK functionality via R. But, it requires MOSEK...

Perhaps others with more refined knowledge will jump in. It might also be reasonable to consider contacting what appears to be a small-ish group of common authors that I came across in the search who have described the methodology (e.g. Taylan or Yerlikaya-Özkurt) to see if they are aware of any R implementations.

Regards,

Marc Schwartz


> On Jan 4, 2019, at 5:16 PM, Bert Gunter <bgunter.4567 using gmail.com> wrote:
> 
> rseek.org might be a better place to search if you haven't tried t
> herealready. However, my minimal effort there did not turn up any R
> software.  Maybe you can do better.
> 
> -- 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, Jan 4, 2019 at 11:31 AM varin sacha via R-help <r-help using r-project.org>
> wrote:
> 
>> Dear R-experts,
>> 
>> We can fit MARS regression using the packages "earth" and/or "mda" or
>> others packages.
>> However, I am wondering if it is possible to fit a CMARS (Conic
>> multivariate adaptive regression splines) using R ?
>> I have googled "conic MARS with R software", I did not get anything, so
>> Google is not my friend anymore !
>> 
>> If you have any solution, would be highly appreciated.
>> 
>> Best,