Data Mining Technology In Credit Card Credit Risk Assessment

21st century is the era where modern information technology advances by unprecedented pace and the development of electronic communication industry and improvement of the handling capacity of computer enormously impacted the traditional money market. After China's entry into the WTO, the money market will open from wholesale to retail market to overall opening and its competition pattern will also change along with that. The foreign commercial banks have rich technological capability and modernized management. They will not only penetrate the Chinese traditional financial businesses but will also aim at the financial intermediate businesses. Credit card business is one of the banks' goals of obtaining profit. This introduces new challenge for the development of credit card business for our banks. Therefore it shall be planned and prepared early in order to strengthen our own capabilities and improve key competitiveness. Compare with the advanced credit risk management system for credit card of foreign banks, our system has great disparity. At present, the credit risk management strategy of credit card business for foreign banks has evolved from simply " preventing risk" to " maximizing profit ". With the development of management idea,the corresponding management tool is progressing too. The risk management tool of foreign banks' credit card business is being developed towards data analysis , setting up of data warehouse , applying data mining technology and various kind of analysis tools like SAS to produce various kind of valuable risks management indexes in order to make decision. The difficulty of credit risk management for credit card is that it has to encourage and facilitate customers' grant loan or overdraw but at the same time eliminate the risk of possible non-performing account. Conventional credit risk management method does not prove to be effectively useful in countering the risks. Nevertheless, with the introduction of data mining technology, it offers a better option with sound fundamental and good predictive mechanism for credit risk management of the credit card business. The data mining technology based on statistical analysis, artificial intelligence and knowledge discovery of databases (KDD) can adhere to the bank's commercial policy and sum up the background records of " good customer " and " bad customer " through collecting and analyzing a large number of behaviors. These include credit and background record of the customers that will cover age, income, sex, condition of the house, marital status, job, education state, etc. and can accurately calculate the expense ability and refund probability of customer group which has the different attribute value hence setting up first line of defense against credit risk: new customer's credit examination and approval. This article touches on analysis of credit risk management of credit card business as the breakthrough point and summarizing the forefathers' achievement of credit risk management. With the input of customer's credit card personal information and historical trend from Agricultural Bank of China; focusing on the research of credit scoring of those applying for credit card. This thesis is mainly divided into four chapters. Chapter one explains the credit risk management of credit card business. It introduces some basic concept of credit card business. It then analyses the necessity of developing the business and risk management the banks shall adopt to counter the main risk they will face: credit risk. Subsequently, it touches on the assessment and approving method of loan and personal credit risk management global banks generally adoptpresently, namely credit scoring model. Lastly, it analyses the advantages and course of design of credit scoring model and scrutinizing the current practice of Chinese commercial banks' credit scoring model. Chapter two touches mainly on the theory of data mining, technology and application. It explains the basic concept and main technology of data mining that includes concept description, asso...