應用資料探勘技術於信用卡黃金級客戶之顧客關係管理

Using Data Mining Techniques to Enhance Customer Relationship Management for Credit Card Gold Customers

陳麗君(Li-Chun Chen)

指導教授 ： 蔡介元

元智大學/工程學院/工業工程與管理學系/碩士(2003年)

摘要

在台灣信用卡（Credit Card）市場的蓬勃發展下，使用信用卡已逐漸成為國人的消費習慣，舉凡食、衣、住、行、育、樂都可以使用信用卡付帳。發卡銀行業者為了維持市場佔有率，不斷地推出各種促銷方案搶攻市場，例如降低申請資格門檻、紅利積點換禮品、降低循環信用利率、刷卡免年費等多項優惠措施。然而這種全面性的大眾行銷方式，不僅成本高昂且效果有限，使得能為業者帶來的利潤亦是有限。有鑑於此，發卡銀行正積極將有限的資源投注在重要的目標客戶群上，以期讓行銷效益更為豐碩，而重要的目標客戶群就是對公司獲利有很大的貢獻，其所帶來的利潤或淨收益通常在80%以上，此種顧客就是所謂的黃金客戶（Gold Customers）。因此，為了獲取新顧客及鞏固既有顧客，如何對黃金客戶作好顧客關係管理（Customer Relationship Management；CRM）已成為銀行業必須面對的首要課題。
過去在信用卡方面的研究多著重在信用卡使用動機之探討、信用卡風險評估、偵測信用卡詐欺和防止信用卡遭盜用等問題，較少有針對獲利性高的黃金級客戶進行特徵組成與消費行為模式之探討。此外，過去研究亦較少著重在不同時間區段的消費模式改變之研究。由於沒有重視黃金客戶在不同時間的消費模式，偵查各個客戶使用此信用卡的情況是否有所變化，造成發卡銀行很難推出適當的產品和服務，以留住獲利性高的顧客。
有鑑於上述兩項問題的重要性，因此本研究以實際銀行為案例，針對信用卡市場其獲利性高的黃金級客戶進行消費行為模式的探討，並開發一個以資料探勘（Data Mining）為基之CRM運作流程。本研究首先將案例公司兩個年度的原始資料匯入至SQL Server，利用SQL語法將不符合黃金客戶要求條件的資料做刪除，並保留適合的資料。為了瞭解原始資料中多且複雜的顧客特徵屬性，我們利用Label SOM方法建構顧客特徵群聚之SOM網路，在完成顧客分群動作後，將各群集中的各個屬性之重要性作排序，以找出較能夠代表此群集之特徵。在決定完重要的人口統計屬性後，我們結合上述找到的重要人口統計屬性與消費者消費行為屬性，作為模糊決策樹（Fuzzy Decision Tree）之輸入資料，以找出一個較佳的黃金客戶分類模式。最後，利用較佳的客戶分類模式，分析不同等級的黃金客戶之規則，同時也瞭解黃金客戶在兩個不同年度消費類型的改變，以提供企業決策者適當的策略與建議。

關鍵字

信用卡 ； 顧客關係管理 ； 資料探勘 ； 自組織映射網路 ； LabelSOM ； 模糊決策樹

並列摘要

The usage of a credit card has gradually changed people''''s consumption behavior in Taiwan. Credit card is used everywhere from food bill, transportation fee, to entertainment purchase. In order to increase market share, every card issuing bank proposes many promotion plans such as decreasing the application criteria, reducing the circulation interest rate and annual fee, and providing bunches of presents, to attract all possible customers. However, comprehensive marketing like this not only cost a lot but limit in its effect. To address the problem, card issuing banks must concentrate on the gold customer who contribute above 80% of the profit for a company. Therefore, it becomes one of the most important topics to operate a customer relationship management (CRM) to consolidate those valuable customers.
Although several researches are initialized to promote better CRM in credit card field, most of them emphasize in the topics of card usage motivation, risk assessment, and fault detection. Few reports are found in discussing the characteristic composition and expenditure behavior of gold customers. In addition, few researches emphasize in studying the pattern changes when time section shifts. Without understanding clearly the behavior changes, card issuing bank is hard to propose a suitable product and service to gold customers. This makes the lost of the important gold customers unavoidable.
To address above two issues, this research develops a data mining model for improving the operation of credit card CRM. This research will focus on the gold customers of high profits and discuss the consumption behavior changes in different time sections. First, credit card transactions of two years are collected from a case company. In order to understand complicated customer characteristic attributes, LabelSOM method is used to construct customer characteristic clusters and find important demographic characteristics of each group. After understanding the importance of demographic attributes, fuzzy decision tree is used to construct a better classification model for gold customers. Finally, we operate several statistic analysis and observe behavior changes in two different years based on the classification model. It is believed that the proposed model can provide an enterprise a better decision-making.

並列關鍵字

Credit Card ； Customer Relationship Management ； Data Mining ； Self-Organizing-Maps ； LabelSOM ； Fuzzy Decision Tree

參考文獻

16.陳來成，「應用資料探勘技術建立商業預測模式－以信用卡為例」，碩士論文，元智大學資訊管理研究所，2002。

27.藍中賢，「結合模糊集合理論與貝氏分類法之資料探勘技術」，碩士論文，私立元智大學資訊研究所，2000。

8.吳國禎，「資料探索在醫學資料庫之應用」，碩士論文，私立中原大學醫學工程研究所，2000。

63.Liu, B., Hsu, W., Han, H. S. and Xia, Y., “Mining Changes for Real-Life Applications,” Second International Conference on Data Warehousing and Knowledge Discovery, 2000, 337-346.

60.Kalakota, R. and Robinson, M., e-Business Roadmap for Success, MA: Addison-Wesley, 1999, 109-135.