search

stan-dev / rstanarm

rstanarm R package for Bayesian applied regression modeling
https://mc-stan.org/rstanarm
GNU General Public License v3.0
382 stars 132 forks source link

Gaussian processes #13

Open jgabry opened 9 years ago

jgabry commented 9 years ago

Not a short-term priority, but could be nice to have some (limited) functionality for GPs in future iterations of the package.

bgoodri commented 9 years ago

Yes, although that would entail writing a bunch of kernel functions, which would better be done in stan-dev/math.

topepo commented 5 years ago

There appears to be some canonical code in the stan docs.

It would be great to formalize this in rstanarm.

dirmeier commented 4 years ago

Hello, are you still planning to include GPs in the package? I currently have some time on my hands and would love to add some functionality, including some kernels to stan/math, in case you're interested.

Cheers, Simon

bgoodri commented 4 years ago

It is a bit difficult to do GPs in a way that would work well for rstanarm users in general. I think we're more likely the approximate GPs like in brms. You can do an approximate GP currently with rstanarm::stan_gamm4 and s(..., bs = "gp").

But adding more GP kernels to stan/math would always be welcome, particularly if there is some family of priors for the hyperparameters that works well for that kernel.

On Sat, Jun 6, 2020 at 4:49 PM Simon Dirmeier notifications@github.com wrote:

Hello, are you still planning to include GPs in the package? I currently have some time on my hands and would love to add them, including some kernels to stan/math.

Cheers, Simon

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/stan-dev/rstanarm/issues/13#issuecomment-640116397, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAZ2XKUXTRVYTWWNGJJHEBDRVKTXFANCNFSM4BNJRXFA .