Chapter 4
Understanding Natures of Uncertainty, Risk Margins and Time Bases for Probabilistic Decision-Making
In order to guide the key choice of the quantity of interest or risk measure, which is the first step in a risk/uncertainty-conscious modelling approach, this chapter discusses in more depth the large variety of choices which may be encountered in practice. These are probabilities (or frequencies) of significant deviations or undesired events, quantiles; more complex level-2 measures (e.g. confidence intervals on failure probabilities); utility expectations; variance of a variable of interest; and any of those probabilistic quantities conditional on deterministic values (or scenarios) for some inputs. The chapter discusses their meaning and properties in more detail, as they:
- may be used to cover differing natures and sources of uncertainty such as intrinsic variability or aleatory randomness, lack of knowledge, imprecision, model inaccuracy, and disagreement between experts (Section 4.1)
- involve differing forms of accounting for the corresponding margins, in particular given the pervasiveness of the monotonous behaviour of the systems at risk (Section 4.2)
- provide different interpretations regarding temporal aspects, being based on what are often implicit stochastic time series; extreme value theory is introduced in that respect (Section 4.3)
- display different decision theory properties that prove more or less helpful according to the salient goals of the study, as defined ...
