在顧客導向的時代,企業需要做好顧客關係管理以便能提昇顧客服務品質與企業競爭力,而要做好顧客關係管理,首先要了解顧客的購買行為。由於顧客購買行為擁有互動性、動態性等多項特色,所以過去用個人經驗來做判斷並無法滿足管理上需求。目前企業都已邁入電子化作業流程,具備蒐集大量顧客交易資料的能力,因此,如何從龐大的顧客資料庫中,挖掘顧客購買行為資訊是項重要的研究議題。 本研究主要目的是要運用多元方式來探討顧客購買行為,藉由食品加工廠公司提供顧客交易記錄,首先,對每一個顧客進行RFM與關聯法則分析,然後結合個人的關聯法則的規則和RFM分數,以k-物件法將顧客進行集群分析,並解釋各集群購買行為特徵。接著,透過k-平均法對顧客分群後再評估價值與先以RFM模型評分後再分群做比較,以了解這兩者的差異。最後,我們以決策樹分析探討RFM模型中R、F與M的權重問題。研究結果顯示,結合規則與數值的關聯法則分析,提供我們對顧客的購買行為有新的瞭解;比較不同辨識顧客價值的方式,提供我們對顧客價值有新的看法;透過決策樹分析,我們了解購買金額(M)在RFM模型中是較為重要的變數。透過了解顧客購買行為與辨識顧客價值,將可以提供企業訂定適當行銷策略,有效進行顧客關係管理,提昇企業競爭力。
In the era of customer orientation, enterprises need to do customer relationship management in order to improve customer service quality and enterprise competitiveness. And to do the customer relationship management, we must first understand customer purchasing behavior. Because the customer purchasing behavior has interactive, dynamic and many other features, to understand it by personal experience cannot meet the management needs. At present, large number of customer transaction data has been gathered and stored in databases, how to analyze the customer purchasing behavior information from large customer databases becomes one of the important research issues. The aim of this research is to investigate the customer purchasing behavior by multivariate approach. First of all, we run k-medoids clustering methods with the combination of association rules and RFM scores data and interpret the meaning of the outcome clusters. Then, the customer groups identified through the k-mean clustering method and traditional RFM scoring method are compared in order to understand the difference between these two methods. Finally, we conduct decision tree analysis to analyze what is the major variable determine the assignment of customer group in previous k-mean clustering analysis. The results of our research show that the combination of association rules and numerical data analysis can provide a new insight of customer purchasing behavior; and compared two different customer values identification methods provides a new perspective of customer value; and through the decision tree analysis, we understand the purchase amount (M) in the RFM model is a more important variable. By understanding the customer purchasing behavior and identify customer value, we will be able to provide enterprise to make appropriate marketing strategies for effective customer relationship management and enhance the competitiveness of enterprises.