BAS: Bayesian Variable Selection and Model Averaging Package

A software package for Bayesian Variable Selection and Model Averaging in linear and generalized linear models, developed by Merlise A. Clyde. It implements stochastic or deterministic sampling without replacement from posterior distributions, supporting various prior distributions and model selection criteria. The work was supported by the National Science Foundation under grant 1106891.

Use Cases

• Performing Bayesian variable selection based on Zellner's g-prior or mixture of g-priors.
• Conducting model averaging for linear models using stochastic sampling algorithms.
• Enforcing model constraints, such as including higher-order interactions only if parent terms are present.
• Comparing model selection criteria like AIC, BIC, and Empirical Bayes estimates of g.

Strengths

• Implements multiple established prior distributions, including Zellner's g-prior and Zellner-Siow Cauchy Priors.
• Supports sampling without replacement and an efficient MCMC algorithm using a tree structure of the model space.
• Allows for user-imposed constraints on model structure, such as forcing variable inclusion.

Limitations

• Row count and dataset scale are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

Merlise A. Clyde

Collection Method

Software package for statistical modeling.