A Language and Program for Complex Bayesian Modelling

TLDR

Gibbs sampling offers powerful analysis of complex data sets, yet its widespread use has been limited by the absence of general‑purpose software, forcing researchers to write custom code in low‑level languages. The authors present BUGS, a general‑purpose software system under development for implementing Gibbs sampling. BUGS consists of a natural‑language model specification interface, an expert system that selects appropriate sampling methods, and a sampling module with numerical routines, all implemented in Modula‑2 and operable on DOS and UNIX, using S objects for data I/O.

Abstract

Gibbs sampling has enormous potential for analysing complex data sets. However, routine use of Gibbs sampling has been hampered by the lack of general purpose software for its implementation. Until now all applications have involved writing one-off computer code in low or intermediate level languages such as C or Fortran. We describe some general purpose software that we are currently developing for implementing Gibbs sampling: BUGS (Bayesian inference using Gibbs sampling). The BUGS system comprises three components: first, a natural language for specifying complex models; second, an 'expert system' for deciding appropriate methods for obtaining samples required by the Gibbs sampler; third, a sampling module containing numerical routines to perform the sampling. S objects are used for data input and output. BUGS is written in Modula-2 and runs under both DOS and UNIX.