The Living Thing / Notebooks :

Bayesian model selection

Apparently frequentist model selection Is not the only type. But what is model selection in a Bayesian context? Surely you don’t ever get some models with zero posterior proability? I thought we just kept all the models and weighted by posterior likelihood when making predictions? I still might wish to discard some models for reasons of computational tractability or what-have-you? Hypothesis testing etc. It’s a bit fraught.


Interesting special case: Bayesian sparsity.