Grappa: R functions for probability propagation
Peter J. Green · 2005
Grappa is a suite of functions in R for calculating marginal and conditional probability distributions on collections of variables with finite state spaces, typically satisfying parsimonious conditional independence assumptions. Typical applications are to expert systems and forensic genetics. It does part of the job to the familiar Hugin system – for a summary of some relative advantages and disadvantages, see Section 4. This document describes use of Grappa; familiarity with using R is assumed. Here is a familiar motivating example: the fictitious ‘Asia’ expert system from Lauritzen and Spiegelhalter (1988), in Grappa: