The Resrearch And Application On Intelligent Decision Technology In Credit Card Approval Process

As a kind of new consumption concept and a kind of new payoff means, credit card has become a product, which many banks are scrambling to launch. The bank credit card brings substantial profits for banks, at the same time there are considerable risks behind the profits. Bank needs to control the credit card risk in the stage of approval with a fix of measure. In previous stage, the approval model is that expert with approval experience checks the new information. Later, banks begin to use the evaluation experts and statistical methods to approval credit card customer. The former approval model is subjective, and the latter didn’t meet the customers’ complexity. Banks bears huge risk. High-profit often coexists with high risk. There are many questions which banks need to be concerned:how to excavate the risk customers from a large number of application information, How to control the risk of bank to the minimum, under the premise of preventing risk, how to obtain the biggest profit.Now, Intelligent Decision Support System has been widely applied to Commercial Processing System. Intelligent Decision Support System based on Data Mining has become a hot area and has got more attentions. In order to deal with these questions which exist in the Credit Card Approval System, this paper make a study of Data Mining, and research the application that classification model in Credit Card Approval System from two factors.Firstly, On Data Mining Classification Algorithm. This paper research and master data mining classification algorithm and research classification algorithm’s classification effect through read large number of documents.Second, this paper compare the classification effect which classification model act on the data in banks. In public data mining system-Weka experiment environment, this paper research data set from Bank of China and find out the most suitable credit card data classification algorithm and classification model. The experiment result shows that the Naive Bayes classification algorithm both in the modeling time and accuracy is superior to the other classification algorithms. Naive Bayes’building model time is shortened0.04s than J48classification algorithms, and is shortened0.3s than SVM.Last, according to the experiment result, this paper applies the Naive Bayes classification model to the banks system which chooses the risk customer. Through building professional Naive Bayes classification model, the paper fully use prior knowledge of old banks customer to judge customer classification.