The Research of Credit Scoring Model about the Commercial Banks’ Customer Based on the Classification First AHP

Abstract: Financial is regarded as the core of modern economy and the banking industry is an important part of the financial industry. Banks' risk control capability is seen as an important component of the core competitiveness. Banks with a comprehensive risk management can compete with the industry to gain the strategic advantage. Therefore, risk management was more and more attention to the bank and various risks become the main threat faced by banks .The risks faced by commercial banks, can be divided into three basic areas. One is credit risk, for example, potential bad debts; one is the risk of liquidity, which would involve the assets and liabilities do not match each other. Another one is the operations risks, such as false individual consumer loans, related enterprises cheat loan, receipt fraud, and so on. Credit risk is the greatest risk of commercial banks which mainly facing to the credit business. It means the possibility that the borrower is unable to perform the contract to repay bank loans on time. Along with the diversity of banking business, credit business, similar to loans, such as discount, overdrafts, letters of credit, guarantees and other businesses of the risks regarded as the control range of credit risk. Credit risk management is the first and key step in a commercial bank's credit business, it means the survival of banks and stability of the society. In recent years, with the reform of the financial system and the acceleration of the opening the financial in China, domestic banks are facing a serious challenge of international competition. Faced to the increasingly globalized financial industry to the new situation, how to strengthen China's commercial bank's credit risk management, and narrow the gap with foreign counterparts, has become a pressing task.The purpose of the article is using DataMining and software to analyse the data source and using the AHP method to rank the credit of customer.The original clients’data are not evaluated nor sorted. The source of data used in the article was from Deutsche Bank. We firstly used decision tree C4.5 to sort the data source, in order to decide the credit status based on the property of each data. We didn’t sort the bad credit evaluation into detail. Those clients with good credit evaluation, we subdivided them using k-means cluster; clients with moderate credit evaluation, we planed to cluster them according their property described situation, into 3 clusters. We analyzed these 3 clusters and established specific evaluation standard for each index and the judgment matrix suitable for the cluster, then confirm the index weight and finally established the credit behavior evaluation model.In general, the innovation of the article is summarized as below:The research angel is innovated. Aimed at the personal clients’credit risk of China’s commercial banks, the article chose the personal clients’credit evaluation as the study object from the banks’point of view, mainly studied the commercial banks’evasion strategy against personal clients’credit risk.Innovation of the AHP method. The article combined data digging technique with traditional AHP, screened and analyzed the clients’data in multi-level. This method then avoided the universal standardization of the data, thus the analysis was more precise and practical, and could effectively help the banks to make reasonable decisions.Embedded scientific quantitative method into the decision-making system of the banks. The data mining technique took the credit analysis to a more scientific and precise level.The modified AHP method, as known as classification first AHP, sorted the source of the dada, then decided the weighted standard of different clusters according to their specific situations. For instance, age group, individuals over 40-years old and below 30-years old, the weight on education status should be different. Individuals over 40-years old generally have lower education level than those below 30-years old, but their economic endurances have significant difference, such situation should be sorted and then evaluated. This only an example, the real calculation and theory are more complex.As well as data mining technique, the elements of data were also analyzed using multi-level analysis, and each property of the data were evaluated. The clients’credit evaluation can then have specific mark, so the banks would make more scientific decisions.Through our research, we have gained a primary understanding on the personal clients’credit risk of the commercial banks in China, provided a quantitative and qualitative method for risk management, evaluated the personal clients’credit by modified AHP, and established an evaluation model using software, technique and knowledge such as WEKA and MATLAB, thus we have a more profound understanding and a clearer frame on credit evaluation. But it must be pointed out that the article only established an elementary and simplified model, more discussion is needed to precisely research the personal client’s credit evaluation, and it must be consummated and developed during practice…
Key words: commercial bank; credit evaluation; data mining; decision tree; k-means cluster; AHP; WEKA