在網際網路日益普及下，近年來航空公司多將線上購票視為網站的必備功能，也因此與旅客之間出現新的互動關係。拜電子商務蓬勃發展之賜，航空公司透過旅客的線上購票行為，能同步獲得豐富的顧客資料及交易紀錄，而如何善用此原始資料，並透過適當的資料處理技術，提供給顧客個人化的行銷及服務，將是未來航空公司的重要課題。本研究應用資料採礦中的群集化、分類及關聯分析技術，對線上購票旅客進行探討，首先用 RFM指標 (recency、frequency、monetary) 及艙等和平均哩程數等五項顧客價值指標對顧客進行群集化，接著以其結果建構顧客分類預測模式，最後則針對不同旅次目的進行航線之間的關聯分析。本研究成果在實地訪談航空公司相關人員後，也獲得實務上的驗證。

In recent years, more and more airlines regard on-line booking as the basic function of their homepage. Since the Internet becomes more popular day by day, the traditional travel agencies are facing the crisis of having the middleman removed. However, a new interaction has appeared between airlines and passengers. While the e-commerce develops vigorously, the airlines go through passengers' online booking behavior, namely to obtain more customer information and transaction records. Therefore, how to use this data to understand the customers, experience suitable data processing technology, and provide the customized marketing service to riders, all have become issues of future airlines. This research will explore data mining to discuss airline passengers’ online booking behaviors. First, we adopt the RFM model (Recency, Frequency, Monetary), the average mileage and classes as five customer value index items to process clustering for riders. The result will make a classification and quickly distinguish customer group belongings. Again, we aim at service products classification to understand the consumers’ behavior of each route. Finally, the association analysis is carried out for different trip purposes, thence appears customers’ implicit connecting demands between all routes.