Estimation of the Accuracy of Methods for Calculating Interval Weight Vectors based on Interval Multiplicative Preference Relations

One of the problems nowadays in solving decision-making problems with pairwise comparison judgements is the choice of prioritization method. The pairwise judgements about preference values of compared objects are given by an expert in conditions of uncertainty, ambiguity and indeterminacy. The paper deals with interval multiplicative preference relations (IMPRs) as one of the popular type of the relations which have been used for representing the pairwise comparison results in Analytic Hierarchy and Network Prosesses. This study attempts to address the key issues of IMPRs: consistency, indeterminacy and the choice of prioritization method. Several experiments were performed with IMPRs of various consistency and in-determinancy level. In each experiment, the mean logarithmic absolute deviation, the mean squared deviation, the root mean square deviation and the difference ratio were calculated for interval weights obtained by six different prioritization methods. The best prioritization method in terms of mentioned four accuracy criteria was chosen for each of the examined IMPRs. The choice of prioritization method should be done for each input IMPR separately. Comparative studies are furnished to demonstrate how resulting rankings by different prioritization methods depend on consistency and indeterminacy level of a given IMPR.