I built a PyMC model that predicts and successfully fits a 2D histogram to an observed 2D histogram with a Poisson likelihood using the NUTS sampler: z_obs = pm.Poisson('z_obs', mu=hist2d_model_mean, observed=hist2d_data) I am able to use the resulting arviz.InferenceData make trace plots, pair pl…

You need to recreate the model, as the InferenceData does not contain information about what the model / likelihood looked like.

How to evaluate Poisson log-likelihood from arviz.InferenceData after successful PyMC fitting?

ricardoV94 May 6, 2023, 8:54pm 2

1 Like

Topic		Replies	Views
"log_likelihood" not found in InferenceData Questions	5	4369	June 21, 2023
How to retrieve (log) posterior/prior/likelihood from trace? Questions	4	4359	May 23, 2020
Outputting loglikelihood of each parameter set Questions	10	685	March 22, 2022
Blackbox likelihood: Creating an Inference Data object and “Initial evaluation of model at starting point failed” Questions	6	1410	February 10, 2022
PyMC outputs loglikelihood slowly using custom models Questions	0	371	November 18, 2021