ساخت حسابرسی اعتبار و مدل کنترل و مدیریت به روش داده کاوی

The 2008 financial tsunami, hitting the globe across all types of industries, causing tides of bankruptcies and severe unemployment, had its epicenter at American subprime in the housing market. In fact, the US subprime storm was just a premonition, while the root cause of the financial tsunami lied in the oversupply of structured credit products. Credit card business, one of the structured credit products, which under an intensively competitive environment, have been released by many banks with high spread, high return, and easy-to-apply appeals to carter to consumers needs. In order to allure the customers, some banks even go to the extent as simplify the credit rating, which in turn has increased credit risk, causing high non-performing ratio, increased debt collection cost, and growing bad debt counts. Accordingly, credit risk auditing plays a vital role in the successful management of credit card business. In response to such needs, the present study aims to conduct analysis and investigation on the current status of the industry with CRISP-DM model. First, customers’ demographic data and payment-related statistics were analyzed to identify feature variables, which were then sorted out as demographic data, debt data, payment rating etc. Next, by utilizing artificial neural network of data mining technique, the study tries to predict customer’s regular pattern of consumption, payment and/or default and bad debt, and to develop a set of credit granting principle by employing the decision tree technique. Since data mining classification model has a greater power in discriminating credit card granting, it can thus be used to construct accurate credit variable rules and predictive model, to further improve credit checking effect and credit risk control. Using the credit auditing data of a certain bank as a case study, the study intends to verify that the model constructed by the researcher can effectively identify the potential key factors of its credit card granting rule, to minimize the cost loss of Model I and Model II credit business, and eventually enhance the stability and profitability of the bank’s credit card business.

To date, credit card has become one of the indispensible market phenomena of currency transaction system. This is particularly true of the world with a predominance of commercial interests, where either on the bank end or the consumers end, a finance management idea of “Enjoy first, pay later” is encouraged. Taking the department stores as an example, almost all of them offer interest-free installment at the anniversary day to attract the consumers to conduct advance consumption with their credit. The shadow of the duel-card (credit card and cash card) storm triggered 3 years ago, still lingers in the public mind. According to the latest statistics, until October 2008, total circulation of domestic credit cards amounts to 36.4 millions. And, this is only the new low since the 2005 credit card bad debt storm, because the number of newly issued cards were outnumbered by the cut and suspended cards. Prior to the explosion of duel-card issue, domestic credit card market had experienced its peak period. 2005 set the historic record high of 45.49 million credit cards, which in 2006, dropped to a circulation of 38.32 million due to the impact of credit card bad debt. Since its official emergence in September 2005, total write-off of bad debt by various credit card issuing banks amounted to NT$13.4 billions. According to statistics released by the Financial Supervisory Commission, Executive Yuan in February 2006, sum total of the circulation interests of credit card and the granted loan balance of cash card had climbed to NT$76.49 billions, with 520 thousand overdue card holders, averaging default payment was NT$300 thousands per capita. Oversupply of structured credit products and over expansion of credit have planted the root cause for the financial tsunami, created the most devastating global-scale financial crisis in nearly 50 years. Such incident makes it imperative that the banking industry should reexamine the way they judge and review the applicants’ credits. Excessive issuance and overdue payments of credit cards have caused grave economic problems. Excessive use by card holders is not the sole cause for credit card problem. A more serious cause is the simplified credit rating among other reviewing processes that the bank used to win over customers in a competitive environment. This has created tens of billions of bad debt by the excessive consumption of insolvent card holders. Accordingly, financial banks, in conducting credit granting, should adopt a set of standards, stringent reviewing mechanism, and on the basis of revenues, try to make the right selection, to minimize the occurrence of bad debt, and to enhance the management performance of card issuance banks. A majority of previous literature focused on constructing credit card classification or behavior classification model with high accuracy, without taking into account the Model I and Model II errors resulted from misclassification. Here, Model I error refers to the misjudgment of applicants with good credit for high-risk group. Conversely, Model II error happens when applicants with bad credit are misclassified as low-risk group. As shown in the following Table 1:In this study, the researcher intends to use the CRISP-DM 6-step cycle of improvement procedure to identify the influential factors causing default discrimination control in the reviewing process. Furthermore, by applying artificial neural network (ANN) and rule of decision tree, to cut down misjudged credit reviewing that cause bad debt resulted from credit expansion, and hopefully, to establish a set of relevant rules that can effectively eliminate those errors.

With its effective prediction of the probability of applicant’s future default, the credit scoring model not only can greatly upgrade the handling efficiency of consumer finance, but can also effectively minimize the credit risk encountered the bank. The bank’s profits and loss depends on the quality of its credit granting system. However, the current credit granting policy popularly adopted by most banks relies most on a scoring method as the guideline, which has been proven weak in constructing accurate credit granting decision-making, and worse even, it has neglected the trend of environmental changes. By combining the ANN and the rules of CART decision tree, the study has effectively reduced the errors of Model I and Model II to 3.4% and 4.7% respectively, and brought up the forecast accuracy of the entire credit granting to as high as 96.5%. As shown in the experimental analysis model, there are four factors that have a significant impact on the applicant’s default probability: (1) Annual income, with an annual income of NT$300 thousands as the baseline; (2) Seniority, with 10-year work history as the baseline; (3) Credit granting and real estate status, with ownership/non-ownership of private house as the baseline; and (4) Family condition, with married/single as the baseline. In making credit checking, the bank should be particularly cautious against the item (s) that has any probability of default. Applicants with more than two possible default items are the high bad debt rate cohort, and should be denied request for card granting. The above rules summed up in the study may provide the frontline auditors a valuable reference to credit banking auditing business, so that their credit scoring sheet may become a more accurate decision-making tool.