Bayes linear graphical models in the design of optimal test strategies

Testing plays a vital role in reducing uncertainty but is resource intensive and identifying the best design is a difficult process. During the development of a system there are a number of potential tests that can be performed with varying efficacy and resource requirements. In this paper we propose a Bayesian modelling process which takes the form of a Bayesian Belief Network (BBN) to determine test efficacy and permits programme managers to assess optimal trade-offs between uncertainty reduction and resources. Supporting inference from a full Bayesian model can be prohibitively expensive computationally so we utilise a Bayes linear approximation, known as a Bayes linear Bayes graphical model, to the inference.