running-bayesian-workflow

What problem does it solve?

It helps you fit Bayesian regression and hierarchical models without shipping unreliable results, by forcing a disciplined workflow for priors, sampling, diagnostics, and uncertainty reporting.

Core Features & Use Cases

• Prior discipline: Choose weakly informative priors that are scaled to the data instead of vague defaults that cause funnels and divergences.
• End-to-end validation: Run prior-predictive checks, NUTS sampling, posterior-predictive checks, and model comparison with LOO or WAIC.
• Diagnostic gatekeeping: Stop interpretation until r_hat, ESS bulk, ESS tail, divergences, and E-BFMI all pass threshold.
• Use case: A data scientist fitting a hierarchical logistic regression in PyMC can use this Skill to fix divergences, reparameterize safely, and report credible intervals only after the fit is clean.

Quick Start

Use the running-bayesian-workflow skill to diagnose my Bayesian model, tighten the priors, verify the sampling diagnostics, and report credible intervals only after the posterior passes all checks.