The New Algorithm for Effective Reducing the Number of Pairwise Comparisons in the Decision Support Methods

The pairwise comparison matrix is often used in different multi-criteria decision-making methods. The matrix identification process is related to the curse of dimensionality because as the size of the matrix increases, the number of queries to experts increases exponentially. One such method is a Characteristic Objects METhod (COMET) which requires building a Matrix of Expert Judgments (MEJ) in order to evaluate alternatives. However, to build MEJ, a decision maker or an expert should answer multiple pairwise comparison questions, which can be expensive in both financial and time terms. In this paper, we propose an algorithm that efficiently reduces the number of pairwise comparison questions to an expert. The algorithm is based on the triads’ consistency and achieves more than ninety percent reduction in questions for big expert judgment matrices. Although the original algorithm is proposed for the COMET method, it could be generalized and used with such methods as RANking COMparison (RANCOM) and Analytic Hierarchy Process (AHP). The efficiency of the algorithm is proved using a simulation approach. The results of those simulations, as well as the pseudocodes, are presented in the paper.