search

PumasAI-Labs / Bayesian-Benchmarks

Code for the Pumas Bayesian Benchmarks Paper
0 stars 0 forks source link

Bayesian VPC #10

Open mohamed82008 opened 2 years ago

mohamed82008 commented 2 years ago

We can do simulations from the prior or posterior predictive distribution and then do a visual predictive check (VPC) plot.

cbdavis33 commented 2 years ago

I have a couple versions of visual predictive checks in R using the Stan/Torsten output. I'm in general not a huge fan of the Upsalla (and now industry standard) VPC plots, so I haven't thought about how to replicate those, but that might be necessary, since that's what people want to see

storopoli commented 2 years ago

We use this following approach:

In Pumas, for models with continuous derived variables, the non-parametric quantile regression approach discussed in Jamsen et al. is used. Discrete and Time to Event models are also supported through the vpc function and can be used without any change in syntax.

So we use a quantile regression. Is this the "Upsalla" standard?

mohamed82008 commented 2 years ago

In Pumas it will be as simple as replacing ... with a few arguments in the following 3 liner:

sims = simobs(...)
vpc_out = vpc(...; simulations = sims)
vpc_plot(vpc_out)