Methods for representing uncertainty

Abstract

This document provides a critical review of different frameworks for uncertainty analysis, in a risk analysis context: classical probabilistic analysis, imprecise probability (interval analysis), probability bound analysis, evidence theory, and possibility theory.

The driver of the critical analysis is the decision-making process and the need to feed it with representative information derived from the risk assessment, to robustly support the decision. Technical details of the different frameworks are exposed only to the extent necessary to analyze and judge how these contribute to the communication of risk and the representation of the associated uncertainties to decision-makers, in the typical settings of high-consequence risk analysis of complex systems with limited knowledge on their behaviour.