應用資料探勘方法中的分群分析技術來探究高階健檢客戶之型態組成以進行客戶關係管理－以台北某醫學中心之高階健檢客戶為例

Consumer Relationship Management Using Cluster Analysis of High Level Health Examination Customer Groups as a Data Mining Technique-Example of a Single Medical Center in Taipei

王建智(Chien-Chih Wang) ； 陳銘樹(Ming-Shu Chen) ； 徐正容(Cheng-Jung Hsu)

《健康管理學刊》 6卷1期 (2008/06) Pp. 49-59

https://doi.org/10.29805/JHM.200806.0005

摘要

隨著人口老化與經濟M型化，預防醫學的觀念日趨受到重視，加上目前各大醫學中心均極力發展非健保給付的自費醫療，高階健康檢查或健康管理服務已經蔚為潮流。目前各大醫院均開始致力於研究目標族群的組成並針對其需求發展出適切的服務，哪些族群應屬於高階健檢之目標客群乃是件極為重要的事。透過科學的資訊工具與方法找出真正的目標客群，研究其真正需求才能進一步做到客戶關係管理。本研究利用資料探勘的分群技術對某醫學中心2005~2006年共計1127筆消費金額超過一萬元的高階健檢客戶所填寫的滿意度問卷，針對人口學變項、消費動機與滿意度資料進行分群分析。分群方法採用二階段法，第一階段以華德法作分群，決定群組個數，第二階段再以K-means法進行群集。透過分群技術進行研究樣本分析後，該院高階健檢主要呈現出三種型態的客群，第一型相似分數為40.42%，該群為女性年紀大約在50~64歲，教育程度在高中職。第二型相似分數為38.80%，該群為男性年紀大約在35~49歲，教育程度在大學且多為公司主管或幹部。第三型相似分數為20.78%，該群為男性年紀大約在50~64歲，教育程度在大學以上且多為公司高階主管。再了解主要客群的組成後，加上其對該院健檢滿意度與消費動機，本研究在結論即訂出相關的顧客關係管理策略。

關鍵字

資料探勘 ； 顧客關係管理 ； 分群分析 ； 高階健檢

並列摘要

The evolution of population aging and M-shaped society lead to a wide acceptance of preventive medicine. Added by the enthusiasm of self-paid medical service market among leading medical centers, high level health examination has become a modern popularity. For the development of customer-oriented medical services, major hospitals are making efforts to analyze the compositional profile of customers. The qualification of customer groups of high level health examination has become one of the greatest interests. A scientific and technical approach is necessary for customer group characterization and customer relationship management. Cluster analysis technique is utilized for data mining. The sample comprised of 1127 questionnaires in a customer service satisfaction survey on health examination customers with payment over NT$ 10,000 dollars from 2005 to 2006 in a single medical center. Demographic features, consumer motivation, and satisfaction are variables for cluster analysis. A two-way clustering procedure is employed. The first step is cluster size determination by Ward’s method, followed by the second step of K-means clustering. The cluster analysis of the sample data yields three classes of the high level examination customers. Customers in the class I, with class resemblance score of 40.42%, are females, typically aged 50 to 64 years, and educations of high school or vocational high school. The class II, with class resemblance score of 38.80%, is composed of males, aged 35 to 49 years, with educations of university, and often managers and major employees. The class III, with class resemblance score of 20.78%, is composed of males, aged 50 to 64 years, educations of at least university, and mostly managers and executive employees. The clarification of major customer partitions is the basis of further analysis of health examination satisfaction and consumer motivation factors, as well as the final goal of hospital policy on customer relationship management.

並列關鍵字

Data Mining ； Customer Relationship Management ； Cluster analysis ； High Level Health Examination