| Article ID: | iaor2002782 |
| Country: | United Kingdom |
| Volume: | 5 |
| Issue: | 1 |
| Start Page Number: | 63 |
| End Page Number: | 75 |
| Publication Date: | Jan 1993 |
| Journal: | IMA Journal of Mathematics Applied in Business and Industry |
| Authors: | Fogarty Terence C., Ireson Neil S. |
| Keywords: | credit scoring, genetic algorithms |
A machine-learning approach that uses the genetic algorithm to optimize a Bayesian classfier from a set of examples is introduced. The IDIOMS system – which has been built to support this approach as well as to allow the incorporation of human expertise into the decision process – is described, and results of using the approach in a tenfold cross-validation comparison with other machine-learning approaches on 51,023 examples of credit-card applications are presented.