Skip to Main Content
This paper presents a simulated annealing based rule extraction algorithm (SAREA) for credit scoring problems. In previous studies, several classification algorithms like statistical models, mathematical programming, and artificial intelligence techniques have been used. This paper aims to illustrate the ability of SA to develop accurate classifiers for credit scoring problems. The use of SA is a new attempt to effectively explore the large search space usually associated with classification problems, a nd to find the optimal set of 'if-then' rules. Experiments are performed on a German Credit Approval data set. We compare SAREA with some classical methods. The results indicate that the results achieved by proposed SAREA are competitive.