The meaningful mixed data TOPSIS (TOPSIS-MMD) method and its application in supplier selection

Abstract

The TOPSIS method suffers from two major shortcomings (1) the non-meaningfulness of the resulting rankings in mixed data contexts (i.e., the rankings of alternatives may change under admissible transformations of the initial attribute values, in the measurement-theoretic sense of the term), and (2) rank reversals or ranking irregularities(i.e., the rankings of alternatives may change if a new alternative is added to the given offered set of alternatives or an old one is deleted from it or replaced in it). The present research tackles the above shortcomings in order to improve the TOPSIS method by suggesting novel reference points and by extending it to mixed data in a rather defensible manner. Finally, the suggested TOPSIS method (referred to herein as the meaningful mixed data TOPSIS method (TOPSIS-MMD)) is used to solve a mixed data multi-attribute supplier selection problem.