銀行業的交易資料庫蘊含豐富的未知知識。例如存款系統的資料挖掘可以了解客戶的資金使用狀況，信用卡客戶交易系統的資料挖掘可以了解顧客的分群與消費行為，進而提供客戶最佳的顧客服務。銀行業長久以來致力於發展日常交易資料庫，但隨著交易類型日趨複雜，交易資料量也隨之迅速擴張，卻未真正達成深入客戶需求，找出客戶分群的狀況，進而著手執行有效的直接客戶行銷與管理。效率對於大型資料庫的知識挖掘這項工作是十分重要的，因為商業資料庫裡通常有大量的資料。而在資料挖掘的演算法中，一個很重要的問題就是如何在大型資料庫裡找出我們需要的資料。本篇論文以銀行大型資料庫為例，結合資料庫與結構化查詢語言，應用於資料挖掘演算法中決策樹的建構過程，包括資料前處理與資料挑選過程，對銀行信用卡資料作有效分群，並導出信用卡客戶分群規則，以提供信用卡審核的信用評等之用，另外，也對資料量多寡對於演算過程效能作分析。

Transaction databases in the banking industry contain a large amount of unknown knowledge. For example, applying data mining to the deposit system can realize the financial usage of customers. Besides, applying data mining to the credit card system can find the classification of customers and realize their consuming behaviors. As a result, a bank can utilize this information to provide the best service to its customers. For a long time, the banking industry puts a lot of efforts in developing online transaction processing systems. The types of transactions are becoming more and more complex and the amount of data is also growing up rapidly. However, the bank manager still fails to meet the customer's need and does not find the customer classification status to carry on an efficient direct marketing and customer relationship management. Efficiency is very important in data mining, particularly for large databases. Nevertheless, commercial databases always have a large amount of data. Therefore, a very important problem in data mining is to find required information from large databases. This paper uses a large bank database as an example. We integrate database technology with SQL and apply it to the decision tree building process, which includes the data preprocessing and data selecting processing. This process can effectively clas-sify credit card data and induct classification rules of credit card customers, so as to support credit scoring in personal credit assessment. In additional, we do an analysis between data amount and decision tree building process performance.