The Comparison of Interval Priority Weight Estimation Methods Using Three Representative Solutions under Various Settings of Vague Preferences

Abstract

In the interval analytic hierarchy process, interval priority weights are estimated from the given pairwise comparison matrix to reflect the vagueness of human evaluation. The solution to the estimation problem is usually nonunique, and the solution set frequently becomes a line segment connecting two interval priority weights. For the performance evaluation of an estimation method for interval priority weights, the consideration of the whole solution set is preferable. In this paper, several estimation methods for interval priority weights are compared in terms of their performance in ordering alternatives. Although the performances have been evaluated under five particular settings of preferences, the results can depend on the settings. In this paper, under the general settings of preferences generated by random numbers, the performances in ordering alternatives are evaluated by using standard, minimum, and maximum solutions.