Dealing with redundancies among criteria in multicriteria decision making through independent component analysis

Redundancies among criteria are frequently observed in multicriteria decision making problems. This aspect may introduce bias in the achieved ranking, since alternatives that have a good performance only in redundant criteria will be favored in comparison with other ones. Aiming at dealing with this inconvenience, several sophisticated approaches were developed in the literature. Most of them require subjective information provided by the decision maker in order to define the set of parameters. In this paper, we consider one of the simplest and most used method, called weighted arithmetic mean, and propose some approaches to automatically adjust the weights in order to deal with redundant criteria. For this purpose, we formulate the problem as a blind source separation one and apply independent component analysis techniques to extract the information used to perform the adjustments. Numerical experiments in both synthetic and real data attest our proposal. Moreover, we extend the proposed approaches to deal with group multicriteria decision making problems.