BayesMix: An R Package for Bayesian Mixture Modeling

BayesMix implements Bayesian estimation of finite mixtures of univariate Gaussian distributions with MCMC methods. Within a given model class the user can modify the prior specifications and the initial values for developing a suitable model for his data. Furthermore, tools for analyzing the output of the MCMC simulations as, e.g., diagnostic plots are available. BayesMix uses package rjags for the MCMC sampling.

Package rjags builds on JAGS (Just another Gibbs Sampler). JAGS is a program for analysis of Bayesian graphical models with Gibbs Sampling and reimplements the functionality from WinBUGS. It is written by Martyn Plummer. The JAGS sources together with a manual as well as Windows binaries can be downloaded from here.

The package was presented at the first R user conference useR! 2004. The submitted abstract and the poster provide further information. The vignette gives a short introduction and some examples.