A new method for credit estimation based on improved objective cluster analysis (OCA)

Credit data, the data that describes the attributes of customer credit collected by enterprises or institutions, which contains a wealth of credit information, is the important basis of customer credit scoring. Using data mining technology to analyze the credit data and evaluate credit of customer has become a highly efficient method for customer credit estimation. Related research has become a hot issue in recent years, and the clustering methods have also got good application and development in these field. This paper briefly describes the present main research methods of the customer credit estimation, and introduces the theory of group method of data handling (GMDH) and the objective cluster analysis (OCA) method. By analyzing the OCA method used in customer credit evaluation on the advantages and problems, this paper introduces the Mahalanobis distance and the combination criteria to control the fitting precision and consistency, and solves the condition of multiple solutions which is common in traditional OCA method. Finally, the validity of the method is verified by testing the actual data and comparing with traditional OCA method.