模化航空旅客預先購票行為

MODELING ADVANCE PURCHASE BEHAVIORS OF AIR PASSENGERS

劉佳欣(Chia-Hsin Liu)

指導教授 ： 邱裕鈞

國立交通大學/管理學院/運輸與物流管理學系/博士(2016年)

摘要

模化航空旅客預先購票型態 (advance purchase patterns) 是了解航空旅客行為之關鍵，並可做為航空公司定價、產品銷售、行銷企劃與航班調度等策略規劃之依據。航空旅客預先購票行為 (advance purchase behaviors) 是複雜且多變的，受到許多因素如機票動態訂價、航空公司競爭、旅客行程規劃、以及旅遊季節性影響。回顧以往文獻，針對旅客預先購票行為之研究，大多以發放調查問卷之方式進行探討。然問卷調查之研究方法有調查成本高、費時，以及回收樣本數有限之缺點。近年來，航空公司營收於線上售票之占比逐年上升，航空營收管理策略與旅客之間互動行為也越趨複雜。因此，與傳統問卷調查研究方法相比較，透過售票歷史資料進行航空旅客行為之研究與預測，不僅更為直接、成本低廉，且更具有代表性。據此，本研究收集 2011 年台北澳門航線歷史售票資料，針對預先購票曲線 (advance purchase curves)，以函數資料分析 (functional data analysis, FDA) 探討不同特性航班預先售票曲線之型態與特性，並據以預測特定航班於特定時間之銷售；再進一步以離散多項羅吉特模式 (discrete logit model) 以及連續羅吉特模式 (continuous logit model)，探討旅客出發時間與價格偏好對預先購票行為之影響。本研究之預測結果，可進一步與實時售票狀況比較，提供航空公司機票訂價與銷售策略參考。此外，研究中發現不同特性航班之預先購票曲線，具有相當大之差異。

關鍵字

航空需求 ； 提前購票型態 ； 航空旅客行為 ； 函數型資料分析 ； 函數型同步迴歸分析 ； 離散多項羅吉特模式 ； 連續羅吉特模式

並列摘要

Accurate advance purchase patterns can provide valuable insights into air passengers’ behaviors and can be used to support airline decision-making activities with respect to seat allocation, pricing, marketing and flight scheduling. Advance purchase behaviors are complex and compounded with lots of factors, such as price dynamics, airline competition, trip flexibility, seasonality, etc. Previous studies conduct questionnaire surveys on air passengers so as to develop advance purchase behaviors. The samples collected by questionnaire surveys can represent real individual advance purchase behaviors, but the number of valid samples is usually rather limited and required of high survey cost. In contrast, with the growing revenue share of online purchasing, to develop and predict the collective advance purchase behaviors of flights directly based on transaction data is obviously much more timeliness, cost economic, and representative. The predicted advance purchase levels at a specific time of a specific flight based on historical transaction data can be viewed as a reference level in comparing with current sales data so as to dynamically advise pricing and promotion strategies prior to departure. Additionally, according to our analyses on the air ticket transaction data, it is found that advance purchase patterns differ remarkably across flights.

To explore flight advance purchase patterns, a functional concurrent regression model was firstly proposed. Several factors contributing to aggregate advance purchase patterns of various types of flights including flight schedule attributes (such as time of day, day of week, months of year and special vacations) and historical load factors were examined based on the shape of the advance purchase curve of each flight. The ticket transaction data which containing 1,044 flights and 134,820 transaction records of Taipei-Macau (TPEMFM) route in 2011 was used for model estimation. With better learning of advance purchase patterns and passenger behaviors for sales flights, airlines are able to develop and make appropriate adjustments for current strategy more efficiently and compete more effectively in today's marketplace.

Furthermore, the advance purchase behaviors of individual air passengers are considered. As airlines dynamically adjust prices and sales strategy based on the learning sales patterns, passengers can also decide to purchase at the ongoing price or choose to delay their purchase decisions. Therefore, choice models including discrete multinomial logit model and continuous logit model are proposed for the empirical analysis of advance purchase behaviors of air passengers. By modeling both price and departure time preferences of air passengers, the individual choice model developed in this research is expected to offer a rich behavioral interpretation of advance purchase behaviors and allow airlines to evaluate potential impacts of the implementing strategies. The models developed in this research have the potential to both improve existing applications in seat allocation and extend the scope of applications to other areas of airline planning such as pricing and revenue management.

並列關鍵字

Air transport demand ； Advance purchase patterns ； Advance purchase behaviors ； Functional data analysis ； Functional concurrent regression model ； Discrete multinomial logit model ； Continuous logit model

參考文獻

Abou-Zeid, M., Rossi, T. and Gardner, B. 2006. Modeling time-of-day choice in context of tour- and activity-based models. Transportation Research Record 1981, 42–49.

Ben-Akiva, M., Abou-Zeid, M., 2013. Methodological issues in modelling time-of-travel preferences. Transportmetrica A: Transport Science 9(9), 846–859.

Ben-Akiva, M., Litinas, N. and Tsunokawa, K. 1985. Continuous spatial choice: The continuous logit model and distributions of trips and urban densities. Transportation Research Part A: General 19 (2), 119–154.

Bhat, C.R., Steed, J.L. 2002. A continuous-time model of departure time choice for urban shopping trips. Transportation Research Part B: Methodological 36 (3), 207–224.

Bilotkach, V., Gorodnichenko, Y., and Talavera, O., 2010. Are airlines’ price-setting strategies different? Journal of Air Transport Management 16(1), 1–6.