- 學位論文
應用序列樣式探勘技術於行為變化之研究
Applying Sequential Pattern Mining Technologies for Behavior Change Detection
摘要
在以滿足顧客需求為競爭關鍵的時代中，管理者若能有效掌握顧客的行為脈動，適時提供正確的服務滿足顧客的需求，便能提升企業競爭優勢，也因此瞭解顧客行為的變化就成為企業獲利與否的關鍵因素。雖然目前已有很多研究在於探討顧客行為的規則，卻鮮少於討論隨時間變動的行為變化議題。有鑑於此，本研究應用序列樣式探勘技術探討在兩個不同時間區段下的顧客行為變化，並且運用Microsoft SQL Server 2000中的範例證實本研究之可行性。本研究首先應用AprioriAll演算法分別挖掘不同時間下的序列樣式。接著將挖掘出的序列樣式依照序列樣式比對方式計算其間的改變程度，再根據三種不同類型的定義對序列樣式進行分類(即分類為emerging sequential patterns, unexpected sequence changed and added/perished sequential patterns)。最後，再由每種類型中選出顯著改變的行為樣式分析討論，以供為管理著制定策略的參考方針。
並列摘要
To satisfy customer’s requirements and increase competition in market, it is critical for an enterprise to understand changes of customer behavior. If managers can understand changes of customer behavior, they can retain customers through providing appropriate products and services to satisfy their needs. Although many researches have focused on knowing the regularity of customer’s purchase behavior, little attention has been paid to mine change of sequence in databases collected over time. Therefore, the objective of this research is to develop a systematic method to discover the change of customer behavior, and provide an implementation case to demonstrate the feasibility of the proposed method. The proposed method uses sequential pattern mining to explore the change of behavior sequence in different two time-periods. First, the AprioriAll algorithm is used to discover all sequential patterns in different time-periods. Then sequential patterns are clarified as one of three change types (emerging sequential patterns, unexpected sequence changed and added/perished sequential patterns) through proposed sequential pattern matching method to understand the degree of change. Finally, a set of sequential patterns with significant change are retrieved. With the useful information, managers can make better business decision.