To anyone who has installed my rmsb Bayesian modeling package for R: Thinking about dropping support for rstan & using the more up-to-date cmdstan approach. Would you be amenable to installing the cmdstan system executable package on your system? #RStats @mcmc_stan
15
6
59
Absolutely, one thing that would be amazing is putting an option for within chain parallelization like brms. Dr. Harrell, do you have any plans for adding a parametric PH survival model to rmsb?
1
2
Great suggestion. I plan to capitalize on cmdstan's within-chain parallelization in the release following the upcoming one. Regarding survival modeling I'll probably leave that to the amazing rstanarm survival which I wish was on CRAN. @mcmc_stan
1
2
One #RStats issue I need to wrap my head around: cmdstanr is not on CRAN so use of it in my package implies that cmdstanr is listed as a suggested and not an imported package, and no running examples are included in help files.
1
1
1
I had good luck with Suggests:, even couples with the Additional_repositories: line in DESCRIPTION, and proper conditional use behind `if (requireNamespace("pkg", quietly=TRUE))`. That may be an option here as long as `cmdstanr` is at arm's length only.
Dec 2, 2022 · 2:47 PM UTC
1
1


