metrumresearchgroup / bbi

Next generation modeling platform
12 stars 2 forks source link

Read shrinkage for Bayes in NM 7.5 #212

Open timwaterhouse opened 3 years ago

timwaterhouse commented 3 years ago

NM 7.5 now outputs a .shk file for BAYES/NUTS models. Would be nice to have bbi grab this and output in the model summary.

seth127 commented 3 years ago

How urgent is this to get in? I guess the two questions are:

  1. how bad to you need this information?
  2. how easy is it for you to extract manually currently?

We'll definitely do this, just trying to get a sense of when we should prioritize it.

timwaterhouse commented 3 years ago

Not at all urgent. This is something we usually calculate outside NONMEM anyway (after merging results from multiple chains). Would be nice to have for quick evaluations for a single chain, but even then it's fairly trivial to grab from the .shk or calculate. So... lowest of low priorities as far as I can tell.

dpastoor commented 3 years ago

Do we know if the structure is any different than non Bayes this could be as easy as a little extra conditional logic if they are consistent

timwaterhouse commented 3 years ago

Structure looks to be exactly the same for Bayes and non-Bayes for the examples I've tried.