Research, Quantitative Analysis Based On Fuzzy Bayesian Network Credit Card Credit Risk

Credit card business is a high-risk and high return business.Credit card issuers via effective analysis,assessment and measurement of credit risk, to control risk within the reasonable limits,and to gain maximum profit is the pursuit of an eternal theme. But the credit card industry in China is in the early stage of development, to seize market share, many card issuers often overlooked the risk factors, only simply to pursuit the card's quantity and scale, this makes the credit card industry increasing potential risks .In this paper, the research object is the most important and prominent form of risk - credit risk credit risk, we main study the analysis and measurement techniques on credit card credit risk, try to use Bayesian networks with fuzzy set theory in credit risk analysis field. Using the advantages of the Bayesian network management in dealing with uncertainty and imprecision data to enhance the research ability and improve the level of credit risk management of credit card.To quantify the credit risk and to identify the key event of the associated risks, form the angles of credit occuring process and human failures, we use the causal relationship of the credit risks, Combination of qualitative and quantitative analysis, to establish the Bayesian network of the credit risk. In order to facilitate the account ,we use the statistical techniques to gain a tidy Bayesian network in further. Because of the uncertainty of credit risk ,we need combine with the experts'knowledge, to make out the judgement of the basic events'prior probabilities and conditional probabilities. To try to decrease the personal error, we unit the fuzzy set theory, through the fuzzy semantic and integral value to quantify the basic events'conditional probabilities and prior probabilities. By an application for instance, aim at a customer's card information of a commercial bank, with the panel of experts, including conditional probabilities and prior probability ,then reckon the credit risk's occuring probability according as Bayesian network discursion. Further more by sensitivity analysising, we could estimate the key event. In credit card business, the cardholder objectively loss the repayment ability is the key event in causing credit risk. In causing the key event of the top incidents, the economic conditions deteriorate of cardholder is more important. With the bank of thousands of credit card customers, we can use the application of the way to calculate the credit risk occuring probability ,thereby we can categorize the card client to ake risk management.