I was looking for a way to calculate the Bayes risk of brms models. Since I could not find anything, I was wondering, if there is a simple way on how to do this or brms uses a different name.
I am not sure I understand you completely, but you should be able to compute almost any quantity of interest from posterior samples or posterior predictions (
posterior_linpred) - if you know how to compute risk for one posterior sample (one model configuration), then you can compute it for all posterior samples, giving you posterior samples of the quantity of interest.
Does that make sense?
thank you for the hint. I think this can work, but I will need some time to do this. I will come back and provide a solution as soon as I found something.