Correlated Criteria in Decision Models: Recurrent Application of TOPSIS Method

Abstract

The purpose of multicriteria decision models is to help decision maker to evaluate each alternative and to rank them in descending order of performance. A problem can appear when the criteria are not independent. This study explores the effect of multicollinearity between criteria in decision making with the technique for order preference by similarity to ideal solution (TOPSIS) and proposes an algorithm to resolve the problem. The algorithm was based on the application of the TOPSIS method several times until all the components are uncorrelated. The algorithm was applied on two examples from medical field to demonstrate its effectiveness. After we applied the purposed algorithm on two examples the index result from TOPSIS was equal correlated with all the criteria.