Multi-group discrimination using multi-criteria analysis: Illustrations from the field of finance

The primary objective in the discrimination problem is to assign a set of alternatives into predefined classes. During the last two decades several new approaches, such as mathematical programming, neural networks, machine learning, rough sets, multi-criteria decision aid (MCDA), etc., have been proposed to overcome the shortcomings of traditional, statistical and econometric techniques that have dominated this field since the 1930s. This paper focuses on the MCDA approach. A new method to achieve a multi-group discrimination based on an iterative binary segmentation procedure is proposed. Five real world applications from the field of finance (credit cards assessment, country risk evaluation, credit risk assessment, corporate acquisitions, business failure prediction) are used to illustrate the efficiency of the proposed method as opposed to discriminant analysis.