Using classification as an aggregation tool in MCDM

In this paper, we propose a method to obtain the set of best alternatives, when data involved in MCDM are of different types (e.g. numbers, linguistic labels or truth values). This method consists of a classification of the a lternatives according to criteria, and then an ordering of the classes obta ined. In this way, we study each alternative in relation with the other one s. The groups formed indicate that their members are equally preferred over the others. Otherwise, we describe some specialized methods to be used in case that all criteria have the same type of values. (C) 2000 Elsevier Scie nce B.V. All rights reserved.