Masahiro Inuiguchi

Multiple Criteria Decision Analysis via Visualizing Hesitation: An Interval Priority Weight Estimation Approach

Conventional multiple criteria decision analysis typically constructs crisp models from preference data, removing ambiguity in order to simplify the model and produce decisive recommendations. This talk explores an alternative perspective: preserving ambiguity within the decision model so that potential outcomes can be visualized and the robustness of a decision can be evaluated.

The approach is introduced within the analytic hierarchy process (AHP). Instead of estimating a single crisp priority vector from a pairwise comparison matrix, interval priority weights are estimated to represent uncertainty and hesitation in the decision maker's preferences. This produces a set of possible priority-weight vectors and allows multiple potentially recommendable orders of alternatives to be examined.

The talk will present the mathematical formulation of the interval priority weight estimation problem and discuss important properties, including the non-uniqueness of its solutions. Several estimation methods and numerical experiments will be introduced to demonstrate their performance when a decision maker provides vague evaluations of priority weights.

A visualization framework based on the Hurwicz decision rule will then be used to map recommendable orders of alternatives according to different levels of discretion and optimism. A unique order appearing across the map indicates a robust recommendation, while multiple orders reveal the decision maker's hesitation. The resulting visualization enables decision makers to understand how their decision policy affects the final ranking and to make more informed decisions with confidence.