Inducing Valuable rules from Imbalanced Data: the Case of an Iranian Banks Export Loans

problem such as support vector machines, neural networks and rule based classifiers. Rule bases are more favorite in credit decision making because of their ability to explicitly distinguish between good and bad applicants in a credit scoring context, imbalanced data sets frequently occur as the number of good loans in a portfolio is usually much higher than the number of loans that default. This paper explores the suitability of RIPPER, One R, Decision table, PART and C4.5 for loan default prediction rule extraction. A real database of one of Iranian banks export loans is used and, class imbalance issues is investigated in its loan database by randomly Oversampling the minority class of defaulters, and three times under sampling of majority of non-defaulters class. The performance criterion chosen to measure this effect is the area under the receiver operating characteristic curve (AUC), accuracy measure andnumber of rules. Friedman‟s statistic is used to test for significance differences between techniques anddatasets. The results from study show that PART is the best classifier in all of balanced and imbalanceddatasets.