Using Genetic Algorithms for Credit Scoring System Maintenance Functions

Abstract

Due to recent advances in high-speed computing technology sophisticated tools for prediction have been introduced which include genetic algorithms. Genetic algorithms operate with the objective of replicating the manner in which genes are passed from one generation to the next. The surviving genes create models that form the most durable and effective offspring. Due to a high-level of understanding of the domain it is has proved to be very difficult for advanced techniques including genetic algorithms to outperform traditional techniques of credit scoring such as logistic and least squares regression. However, in large organizations it is difficult to continuously update complex scoring systems. This paper explores the use of genetic algorithms to automate and institutionalize the scoring maintenance function within an organization.