A NEW APPROACH BASED ON CLASS IMBALANCE LEARNING FOR SMALL-BUSINESSES' CREDIT ASSESSMENT

摘要

This paper presents a new approach which uses both k-nearest neighbor (k-NN) algorithm and random forest method to deal with the imbalance data sets in a small-business' credit assessment.Two types of classifier are designed.The first one is called as preliminary classifier, which is constructed by using k-means clustering algorithm based on the test data in order to save the useful information of the customers of majority class as much as possible.The second one is called as minority classifier, which is constructed by using random forest method and used to reclassify the customers that were predicted to belong to minority of class in preliminary classification to improve the classification performance of the minority class.The proposed approach is compared with the approaches that use only k-nearest neighbor or random forest method.Through the comparison the advantages of the proposed approach over other methods will be showed.