Can I perform model comparisons using this skill?

Yes, it includes patterns for Leave-One-Out Cross-Validation (LOO-CV), WAIC (Widely Applicable Information Criterion), and DIC (Deviance Information Criterion) for model selection.

Which Bayesian frameworks are supported by this skill?

This skill provides implementation patterns for both Stan (via cmdstanr) and JAGS (via R2jags and coda), covering the most common MCMC engines used in data science.

How does this skill help with Stan divergences?

It provides specific code to identify divergent transitions and recommends common fixes such as increasing the adapt_delta parameter or switching to a non-centered parameterization.

What should I do if my Rhat values are above 1.1?

Rhat values above 1.1 indicate non-convergence. The skill suggests running longer warmup periods, checking for multi-modality, or providing better initial values for your chains.

Bayesian Model Diagnostics

Name: Bayesian Model Diagnostics
Author: choxos

bychoxos

0•

Data Science & ML

Evaluates Bayesian model convergence and sampling performance using MCMC diagnostics for Stan and JAGS frameworks.

This skill provides a robust framework for validating Bayesian statistical models by implementing essential MCMC diagnostics. It guides users through assessing convergence metrics like Rhat and Effective Sample Size (ESS), identifying sampling issues such as divergences and treedepth limits in Stan, and performing posterior predictive checks to ensure model fit. By providing standardized code patterns for both Stan and JAGS, it helps data scientists ensure their posterior estimates are reliable and their models are statistically sound before proceeding to inference.

Key Features

01Visual diagnostic tools like trace plots and rank histograms

02Stan-specific divergence diagnosis and NUTS troubleshooting

030 GitHub stars

04Model comparison via LOO-CV, WAIC, and posterior predictive checks

05Comprehensive MCMC convergence assessment (Rhat, ESS bulk/tail)

06JAGS diagnostic suite including Gelman-Rubin and autocorrelation

Use Cases

01Validating posterior reliability in complex hierarchical Bayesian models

02Comparing multiple model fits using cross-validation and information criteria

03Fixing sampling geometry issues in Stan using reparameterization techniques

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add choxos/bayesianagent model-diagnostics

For use in Claude.ai and ChatGPT

Download Skill