ALGORITMOS GENÉTICOS E MINERAÇÃO DE DADOS APLICADO NA PREDIÇÃO DA INADIPLÊNCIA DE CLIENTES

Abstract

This study addresses the use of genetic algorithms (GAs) in conjunction with Data Mining, to predict whether a customer receiving credit will become delinquent, using the database provided free of charge by researcher Cheng Yeh. The choice of AGs is due to their flexibility and the fact that the data contain noise, insertion errors, variability and patterns are unclear. The quantitative and descriptive methodology was used that will evaluate the performance of the algorithm in relation to other data mining techniques. The results obtained were satisfactory, because the GA can predict with an f-score better than several alternative options, although the performance does not stand out so much in the face of the simplest alternative techniques due to the fact that there are only two possibilities and the proportion between the possibilities is very high.