Credit Assessment Using Evolutionary MLP Networks

Abstract

Credit assessment, in the form of credit card, direct credit to consumer and check card, is usually carried out either empirically or through a credit scoring system based on discriminate or logistical regression analysis. In the past years, a growing number of finance institutions has been looking for new techniques to improve the profit of their services, reducing the delinquency rates. For such, new techniques have been proposed, among them, neural networks. In neural networks design, a few parameters must be adequately set in order to achieve an efficient performance. The setting of these parameters is not a trivial task, since different applications may require different values. The “trial-and-error” or traditional engineering approaches for this task do not guarantee that an optimal set of parameters is found. Recently, genetic algorithms have been used as a heuristic search technique to define these parameters. This article presents some results achieved by using this technique to search optimal neural architectures for credit assessment.