航空公司制定艙位配置與超額訂位策略的良窳，直接影響本身的競爭
優勢。藉由將訂位容量做最有效的管理，對於提高承載率、獲利能力，以
及在市場上的相對競爭力等，皆有相當大的幫助。因此，如何制定最佳的
艙位配置與超額訂位策略是非常重要之營收管理課題。而目前國內航空公
司普遍的作法是以人員經驗進行艙位配置與超額訂位之決策，但此種方式
耗時費力，且需付出相當高之人力成本。故本研究建立一數學模式來有效
地協助航空公司進行艙位配置與超額訂位問題之決策。 本研究係以機
位控制員微觀角度之觀念，將決策單位改以每一艙位，然後依照每一訂位
需求劃分階段，發展新整合艙位配置與超額訂位之動態策略模式。本研究
提出改善過去研究之作法與觀點如后： 1. 提出多費率動態艙位價值
函數，在證明為非遞增函數後，可以相當方便地處理不同費率之單席或多
席訂位需求決策。 2. 引進存留機率之概念，同時考慮到取消訂位與
未報到。在決策樹分析下，發現以往艙位配置會受到存留機率影響，必須
予以重新配置，否則會產生偏誤。並據以提出單席或多席訂位之最高費率
等級不一定恆被接受之觀點，維持各費率間在動態訂位過程中之互相競爭
。 3. 模式中提出考慮第0階段不同費率互相競爭登機之作法，使得超
額訂位可以因應各費率等級訂位容量之變動配置，合理地反應航空公司在
超賣情況下，會由高費率等級依續優先登機之實務措施。 本研究所提
出之整合艙位配置與超額訂位動態策略模式，能夠保留往昔策略之優點，
並且可以同時考慮到取消訂位、未報到、多席訂位、動態艙位價值與超賣
成本等諸多因素，實證分析以國內某航空公司為例，以驗證本策略之有效
性與應用性，俾以提供航空公司實際營運之參考。

The quality of a seat allocation and overbooking strategy
affectsairline companies'' competitive advantage directly. The
decision ofbooking capacity , has a great influence on the load
factor and revenue offlight, and the competitiveness of airline
companies. Therefore, how toconstruct the optimal seat
allocation and overbooking strategies are thusan important issue
on revenue management. Domestic airline companiesoften make
decisions on seat allocation and overbooking based on
theknowledge and intuition of experienced staff. This decision-
makingprocess may not necessarily guarantee optimal revenue, and
it is veryoften time-consuming. To help airlines efficiently and
effectively make abest decision, this research develops a model
to achieve maximumrevenue. This research constructs a new
comprehensive seat allocation andoverbooking dynamic strategy
model from the perspective of spacecontrollers. The model''s
decision variables are the seat, and its dynamicstages separated
by every booking demand. The main contributions ofthis research
will compare favorably with the past work in terms of
thefollowing items: 1. It constructs multiple-fare expected
marginal seat revenue (EMSR)function that can quickly decide the
suitable allocation of single-seat ormultiple-seat demand. 2.
It also applies the concept of survival probabilities that
considerscancellation and no-shows simultaneously. By the
decision-tree analysis,the seat allocation will be reconsidered
by survival probabilities, andthen avoids a wrong allocation.
Base on this concept, the highest fare ofsingle-seat or
multiple-seat demand can not always be accepted.Therefore,
different fares'' demands compete with each other in
dynamicstage. 3. It considers the competition of different
fare classes with eachother in zero stage. Then overbooking
process can include the operationof variable bookings of
multiple-fare class. It better addresses thepractical airline
companies'' operations under overbooking situations. This
research will establish a new comprehensive seat allocation
andoverbooking strategy model. It can remain the advantages of
thetraditional approaches, and considers the following factors,
such ascancellation, no-shows, multiple-seat demand, expected
marginal seatrevenue (EMSR) function and deny-boarding penalty
in one modelsimultaneously. A case study is tested on a domestic
airline company. Itis hoped that this research will contribute
to the better understanding ofcomprehensive seat allocation and
overbooking strategy model.