Opportunity losses due to uncertainty in multiobjective decision problems (Q1069851)

This paper examines the problem of determining measures of loss due to uncertain information in multiobjective decision-making (MODM). When using a single value measure to resolve multiobjective problems such as weighting on an \(\ell_ p\) distance standard decision theoretic methods can be used. Expected opportunity loss (EOL) is then method dependent and does not show the effect of uncertainty on the attainment of each objective. An expected multiobjective opportunity loss (EMOL) is developed. EMOL may be calculated for all feasible points in the decision space. The resultant vector valued measure shows both the loss due to uncertainty and due to inefficient choice of the decision variable. EMOL is a function of the decision variables, the objectives and the probability distributions describing the uncertainties, thus it is independent of the method by which the decision variable values were selected.