What is BSG?

BSG is an open-source, flexible framework for Bayesian statistical modeling and data analysis using Markov chain Monte Carlo (MCMC) methods. It allows users to quickly specify statistical models, fit them using various MCMC samplers, diagnose convergence and sample quality, make statistical inferences, and visualize results.

Some key features of BSG include:

• Intuitive model specification language for defining statistical models
• Support for a wide range of univariate and multivariate continuous, discrete, and mixed data models
• Various MCMC samplers for posterior simulation, including Gibbs, Metropolis-Hastings, HMC, NUTS
• Automated convergence diagnostics and output analysis
• Tools for posterior prediction, model checking, and comparison
• Extensive visualization capabilities for traces, posterior densities, fitted models, and more
• Support for Bayesian workflow like prior predictions and posterior predictive checks
• Fast performance leveraging compiled Stan models
• Scripting interface for ease of use and automation

In summary, BSG makes Bayesian analysis accessible to non-experts while also providing advanced users with a powerful and flexible framework for tackling complex statistical modeling problems across many domains.

Official Links