Title: Optimal consistent approximation of a preference matrix in decision making
Authors: Martin Gavalec; Hana Tomášková
Addresses: Faculty of Informatics and Management, University of Hradec Králové, Rokitanského 62, 50003 Hradec Králové, Czech Republic ' Faculty of Informatics and Management, University of Hradec Králové, Rokitanského 62, 50003 Hradec Králové, Czech Republic
Abstract: A new approach to the inconsistency problem of preference matrices is described. The optimal consistent approximation (OCA) is computed as a consistent matrix with the minimal distance from the given (not necessarily consistent) preference matrix A. The approximation is found by a linear optimisation method under the assumption that A is antisymmetric. Moreover, a modified method (MOCA) is suggested for treating incomplete preference matrices. MOCA can also be used for identifying so-called outliers, i.e., the entries badly influencing the distance of the nearest consistent approximation. Outliers are usually caused by errors in data processing. The methods are illustrated by examples.
Keywords: decision making; incomplete preference matrices; consistent preference matrix; optimisation; optimal consistent approximation; OCA; outliers.
DOI: 10.1504/IJMOR.2016.079810
International Journal of Mathematics in Operational Research, 2016 Vol.9 No.4, pp.520 - 534
Received: 15 Dec 2014
Accepted: 08 May 2015
Published online: 16 Oct 2016 *