Aki Vehtari

It seems the commit https://github.com/stan-dev/bayesplot/commit/17c67cebb1ecd1d573cd2097e111247bd35c4461 added contributing guidelines. Can this issue be closed?

I think these are now in `posterior`. Maybe this issue can be closed?

(Sorry for not checking this 2 years ago!) - `loo_ref_oscale

Here is a demonstration of subsampling LOO with simple random sampling and difference estimator ``` library(rstanarm) library(projpred) library(tidyr) library(ggplot2) data("df_gaussian") dat_gauss

`loo` package has `diff_srs_est()` function, which estimates also the variance (needed for SEs), but the code has some mismatch with the corresponding paper, and the estimated variance can be negative,...

`posterior` package can convert `mcmc.list` objects to posterior objects, `posterior` has the best implementations for rhat's and ess's, and bayesplot works with posterior objects. Closing this one.

Now that cmdstanr is getting `laplace()` method to get draws from the normal approximation, it would be great to tag those draws as not being from MCMC, and by default...

I'd fine with explicit coercion, but if automatic then, for example, couple examples are - MCMC draws of the parameters used to compute parameters of predictive distribution and sampling from...

> Specifically I was wondering whether the tails should be fit on the constrained or unconstrained draws Should be fit on the scale that is reported > and if the...

> Then I suggest that the default be tail = "both" I agree.