Fuzzy-valued choquet integral based utility ranking in the credit scoring problem

Abstract

A linguistic approach based on the fuzzy-valued Choquet integral is proposed to rank the customer loan applicants. Non-additive fuzzy decision models, such as fuzzy-valued Choquet integration, imitate the human behavior and human preferences better than additive utility ranking. The reason for this, is that traditional expected utility theory and Bayesian probability provide tools for decision making under uncertainty in terms of probabilistic decisions, whereas experts perception is based on an imprecise probability for which the use of classical probability theory remains insufficient. In the suggested approach, a Choquet integral with a fuzzy-number valued measure is used. The method is applied to customer loan evaluations for a financial institution to verify expert decisions in parallel to extracting linguistic rules of decision making. The results indicate that the proposed method is successful in ranking the customer loan applications with only four fails in total of 135 applications.