Comparing aggregation methods in large-scale group AHP: Time for the shift to distance-based aggregation. (15th June 2022)

• Record Type: Journal Article
• Authors: Duleba, Szabolcs; Szádoczki, Zsombor
• Is Part Of: Expert systems with applications. Volume 196(2022)
• Publication Date: 2022-06-15
• Subjects: Group AHP -- Priority vector -- Aggregation -- Consensus creation -- Rank correlation -- Compatibility; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2022.116667
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: Euclidean Distance Based and Aitchison Distance Based Aggregation Methods presented. Efficiency examined by 96.000 simulation cases and by a real-world case study. In small dimensional cases both new methods overperform the conventional techniques. In high dimensional (7 to 9) cases Euclidean distance aggregation keeps its primacy. Both proposed methods have low computational time and high applicability. Abstract: This paper aims to compare the efficiency of the conventional aggregation methods and the new, distance-based aggregation techniques in simulated and real-world group AHP cases. For the comparison, we not only applied rank correlation methods, but also examined the compatibility among the individual priority vectors of the group and the created common priority vector in the different consensus creation approaches. Results have shown that in small dimensions, both Euclidean Distance-Based Aggregation Method (EDBAM) and Aitchison Distance-Based Aggregation Method (ADBAM) outperform significantly the conventional techniques. In large dimensions, the dominance of EDBAM remains. Since the computational time of the proposed methods (especially EDBAM) is low and EDBAM maintains its efficiency in large-scale group AHP (proven by 96.000 simulation cases) in every possible dimension within the AHP domain, we can state in case of high number of evaluators, distance-based aggregation is a better approach than the conventional methods.