Article ID: | iaor20062714 |
Country: | Spain |
Volume: | 11 |
Issue: | 1 |
Start Page Number: | 3 |
End Page Number: | 26 |
Publication Date: | May 2006 |
Journal: | Fuzzy Economic Review |
Authors: | Zopounidis Constantin, Doumpos M., Kitsios E. |
Keywords: | fuzzy sets, economics |
Credit cards constitute one of the most common forms of consumer loans. The main purpose of this paper is to apply fuzzy data analysis to the credit scoring problem. A neuro-fuzzy classification technique is compared to the logistic regression approach and novel machine learning algorithms that are currently being investigated as credit scoring methods. The 10-fold cross-validation procedure is performed to analyze the generalization properties and the robustness of the developed models. Neuro-fuzzy classification systems allow for prior knowledge to be imbedded in the analysis and utilize human expertise in the form of fuzzy if/then rules to provide an insight into the reasoning mechanism behind the credit approval/rejection decision. This feature is particularly useful in financial applications such as credit granting, where credit analysts should be in a position to provide an explanation for their decisions.