When should I use non-centered parameterization?

Use non-centered parameterization when your data is sparse, the between-group variance (tau) is small, or when you encounter divergences in your MCMC chains.

How does this help with MCMC divergences?

It provides diagnostic guidance and alternative model parameterizations (like non-centering) specifically designed to resolve geometry-related sampling issues in hierarchical structures.

Does this skill support Stan and JAGS?

Yes, this skill provides implementation patterns and code snippets for both Stan and JAGS probabilistic programming languages.

What is partial pooling in Bayesian modeling?

Partial pooling is a technique that allows group-level estimates to shrink toward the global mean, balancing information between individual groups and the entire population based on sample size and variance.

Bayesian Hierarchical Modeling

Name: Bayesian Hierarchical Modeling
Author: choxos

bychoxos

0•

データサイエンスとML

Implements hierarchical and multilevel Bayesian models with optimized parameterizations for robust statistical inference.

This skill provides specialized guidance for developing hierarchical (multilevel) Bayesian models, addressing complex data structures like nested groups or repeated measurements. It assists users in implementing partial pooling strategies to balance group-level and population-level effects, while offering expert advice on choosing between centered and non-centered parameterizations to avoid sampling issues like divergences. With ready-to-use patterns for Stan and JAGS, it streamlines the process of building sophisticated biostatistical models and diagnosing common MCMC performance problems.

主な機能

01Debugs MCMC divergences and sampling issues in complex models

020 GitHub stars

03Facilitates meta-analysis with study-level variation handling

04Implements partial pooling for nested or grouped data structures

05Provides optimized Stan and JAGS code templates for hierarchical models

06Automates selection between centered and non-centered parameterizations

ユースケース

01Conducting longitudinal studies with repeated measurements on the same patients

02Analyzing student performance data nested within different schools or districts

03Performing Bayesian meta-analysis to synthesize results across multiple clinical trials

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add choxos/biostatagent hierarchical-models

For use in Claude.ai and ChatGPT

主な機能

01Debugs MCMC divergences and sampling issues in complex models

020 GitHub stars

03Facilitates meta-analysis with study-level variation handling

04Implements partial pooling for nested or grouped data structures

05Provides optimized Stan and JAGS code templates for hierarchical models

06Automates selection between centered and non-centered parameterizations

ユースケース

01Conducting longitudinal studies with repeated measurements on the same patients

02Analyzing student performance data nested within different schools or districts

03Performing Bayesian meta-analysis to synthesize results across multiple clinical trials

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add choxos/biostatagent hierarchical-models

For use in Claude.ai and ChatGPT