Commercial Bank Credit Rating Based On Data Mining

Risk management is the key of commercial banks management, also is the most important of healthy development. Deposit and loan business is the most important part of comercial banks, and loan business is the main means of profit. Therefore, as the main risk of commercial banks, credit risk needs active and reasonable response.This paper, using decision tree in data mining classification techniques, longitudinal mi-nes by real commercial bank loan information samples. Through data acquisition, integration, pretreatment, and then build decision tree of the credit risk rating model, and then on this basis, draw loan risk rating, help the commercial bank risk management department predicts loan risks.Seeing the current commercial bank loans business information, data related is large and complex, data of different enterprises is unrelated, and there is no completely same data. Therefore, in view of some limitations of the traditional decision tree algorithm for continuous attributes, mostly do research of commercial bank loans risk forecasting system of financial data discretization method and decision tree generation method:for continuous attributes, using clustering method on attribute data in the basic division, then uses fuzzy set method to solve unfair harsh threshold in dividing continuous data attributes; for new properties with the membership, uses the improved C4.5decision tree classification method, and makes application in banking risk prediction.Using the loan business data of a commercial bank system as source of training data, combining data mining thought and method, by the improved decision tree classification, do the loan risk prediction of a commercial bank loan business. The results show that using the improved method can guarantee the accuracy of prediction in the ratings, and at the same time, make the decision tree results more concise, more draw lessons from a meaning.