Skip to the main content

Original scientific paper

https://doi.org/doi.org/10.2478/bsrj-2024-0015

Selection Procedure of the Approximation Methods for Deriving Priorities: A Case of Inconsistent Pairwise Comparisons

Vesna Čančer orcid id orcid.org/0000-0003-1637-208X ; University of Maribor, Faculty of Economics and Business, Slovenia


Full text: english pdf 373 Kb

page 21-30

downloads: 36

cite


Abstract

Background: When pairwise comparisons are used to express preferences for alternatives or judgments on criteria's importance, several methods can be used to derive priorities in multi-criteria decision-making. In the case of inconsistency, different methods give different results. Objectives: The main goal of this paper is to present the procedure of measuring the accuracy of the selected approximation methods based on pairwise comparisons compared to the priorities obtained by the eigenvalue method. It also aims to illustrate the procedure on the numerical example characterised by acceptable inconsistency. Methods/Approach: The presented procedure is based on a prescriptive approach, the fixed ratio scale, reciprocal pairwise comparison matrices, and consistency ratio. Mean absolute deviation and mean absolute percentage deviation are used to measure accuracy. Results: The first result is the theoretical statement of the priorities’ accuracy measurement procedure. The results of the numerical example characterised by the preferences of strength slight to strong plus show that, on average, the most accurate approximation method is the geometric mean method. Conclusions: The research contributes to the literature on prescriptive approaches to decision-making. The results can show potential users which approximation method to use and lecturers which of them to include in the curriculum portfolio.

Keywords

accuracy; analytic hierarchy process; approximation method; pairwise comparisons; priority; simulation

Hrčak ID:

320830

URI

https://hrcak.srce.hr/320830

Publication date:

22.9.2024.

Visits: 125 *