International Forum on Engineering Decision Making

A concept we call Total Uncertainty will be used to illustrate how we combine both aleatoric (reducible) and epistemic (non-reducible) uncertainties in a given problem to assess the confidence an analyst can have in his/her decisions. In illustrating the utility of our concept of total uncertainty we hypothesize that all uncertainty, including our metric of TU, should scale between two extreme cases of uncertainty, i.e., between the case of no uncertainty and the case of maximum uncertainty. A method to quantify both of these extreme cases is presented. We show a direct relationship between total uncertainty and confidence, the latter being an important notion in decision-making. We illustrate the utility of the TU metric by addressing three simple problems that contain different situations involving the two forms of uncertainty.