A METHOD FOR ELICITING EXPERT JUDGMENT TO ESTIMATE PROBABILITIES IN A BAYESIAN BELIEF NETWORKS

摘要

Bayesian belief networks are proving to be a powerful modeling approach in risk analysis. This paper presents a Bayesian method for eliciting expert judgment using a paired comparison method. The conditional probabilities and covariates that result from the elicitation can be used to populate the Bayesian belief networks. The methodology will be illustrated on sample Bayesian belief network for a submarine pipeline application.