Decision analysis based on fused double exponential possibility distributions

Abstract

In this paper, the possibility grades given by an expert are approximated by double possibility distributions, called upper and lower possibility distributions to characterize uncertainty knowledge from this expert. The upper and lower possibility distributions have some similarities to upper and lower approximations in rough set theory. Multi-source knowledge from multiple experts is represented by a set of exponential possibility distributions. Based on the consistency index defined by the possibility measure of each pair of possibility distributions, a fusion model is proposed to integrate multiple possibility distributions into a new one representing a refined knowledge. As an application, a portfolio selection problem with multiple experts is considered.