A Comparison of Rough Sets and Recursive Partitioning Induction Approach : An Application to Commercial Loans

Daubie, Mickaël [FUCAM] Meskens, Nadine [UCL] Levecq, Philippe [FUCAM]

Credit scoring is the term used to describe methods utilised for classifying applicants for credit into ?good' and ?bad' risk classes. This paper evaluates two induction approaches, rough sets and decision tree, as techniques for classifying credit applicants. Commercial loan analysis, the process of evaluating a company's financial strengths and weakness, has often be tackled through multicriteria methods, statistical procedures or neural networks. However, all of them have drawbacks. On the other hand, inductive learning methods have better knowledge representational structure in the sense that the decision tree or the rough sets can be used to derive production rules. If decision tree have already been used for credit scoring, the rough set approach is rarely utilised in this domain. In this paper, we use production rules obtained on a sample of 1102 business loans in order to compare the classification abilities of the two techniques.