Inference and decision analysis based on imprecise probability and likelihoods

The authors consider the implementation of Dempster-Shafer theory in inference and decision making. Decision analysis must often be implemented in civil engineering applications even though the supporting information is very weak. Imprecise probability seeks to more faithfully represent the uncertainties under these conditions. The Dempster-Shafer (D-S) approach to inference and decision analysis is, retrospectively, an implementation of this concept. D-S theory provides a simple inference scheme which utilizes conventional likelihoods as input. This produces an expected utility interval for each decision alternative. The size of this interval is a reflection of the weakness of the information on which the analysis is based and reduces the ability to distinguish between decision alternatives