Hi, I’d like to estimate a model (and then do some model comparisons) using pymc. The model for generating the predicted probability distribution itself depends on an external module and requires a numpy array as input. So, while I have control over how the likelihood is calculated, I have no contr…

Pointwise log-likelihood for loo/waic with custom "black-box" probability distribution

ricardoV94 December 28, 2023, 8:00am 3

The easiest option is to define a CustomDist with a logp function

Topic		Replies	Views
Explanation of logp and customDist for Blackbox likelihood functions v5	4	432	February 2, 2024
LOO/WAIC metrics from pm.Potential based model? Questions	4	668	October 7, 2018
Using pm.DensityDist and customized likelihood with a black-box function Questions	10	6853	August 29, 2018
How to set up a custom likelihood function for two variables Questions	8	13306	March 9, 2018
Likelihood evaluation and DensityDist Questions	2	1709	February 21, 2018