Build Bayesian statistical models

Build and fit Bayesian hierarchical models with MCMC

When you need quantified uncertainty in a statistical model and can validate priors beforehand.

• · data (predictors, outcomes)
• · prior specifications
• · model structure (coords, dims)

• · InferenceData object
• · posterior samples
• · diagnostics (R-hat, ESS, divergences)

• · Python 3.12+
• · PyMC 6.0.1
• · PyTensor
• · ArviZ

• · data standardized and missing values handled
• · priors validated via prior predictive check

• · divergences indicate posterior complexity
• · low ESS from high posterior correlation
• · multimodality (high R-hat)

• · prior predictive check before fitting
• · posterior predictive check vs observed
• · R-hat < 1.01 and ESS > 400