search

ScottClaessens / coevolve

coevolve R package for Bayesian dynamic coevolutionary models using Stan
GNU General Public License v2.0
7 stars 0 forks source link

Automated prior sensitivity checks via power scaling #32

Open erik-ringen opened 2 months ago

erik-ringen commented 2 months ago

Given that we provide default priors, we should emphasize that they are not necessarily safe to use blindly, and that results can be quite sensitive particularly in small trees. One way we could bring this to user's attention is with automated sensitivity checks using power scaling and importance sampling, as described here:

Kallioinen, N., Paananen, T., Bürkner, P. C., & Vehtari, A. (2024). Detecting and diagnosing prior and likelihood sensitivity with power-scaling. Statistics and Computing, 34(1), 57.

There is now an R package that implements this approach called priorsense: https://n-kall.github.io/priorsense/.

ScottClaessens commented 2 days ago

Interesting! So would such a function take the user's dataset and model, and feed them to the priorsense functions under the hood? Or could we just suggest that users check out this package separately from ours?