The Advantage of Interval Weight Estimation over the Conventional Weight Estimation in AHP in Ranking Alternatives

摘要

In this paper, we investigate the significance of interval weight estimation in the setting of Analytic Hierarchy Process (AHP). We consider several estimation methods for a normalized interval weight vector from a crisp pairwise comparison matrix. They have a desirable property. To avoid the non-uniqueness of the solution, we add an additional constraint, i.e., the sum of centers of interval weights is one. A few ranking methods under interval weights are considered. Numerical experiments are executed to compare the estimation accuracy of ranking alternatives under the assumption that the decision maker has a true interval weight vector. The advantage of interval weight estimation over crisp weight estimation is demonstrated.