| Title | Solving the Dynamic Seat Control Problem with Passenger Choice by a Stochastic Program Model |
| Year | 2019 |
| Degree | Master |
| School | Department of Transportation and Logistics Management College of Management National Chiao Tung University |
| Author | Yu-Kuan KUO |
| Summary | Air transportation has been playing a very important role in the transportation industry. The seat inventory control in revenue management (RM) is a very crucial factor affecting the profitability of airlines. In addition to the differentiation in terms of cabin classes (the first class, business and economy classes), airlines also sub-divide the each cabin class into multiple fare classes based on the various class-dependent restrictions, due to the fact that the numerous air passengers may have different trip purposes and various levels of willingness-to-pay. Thus, how to sell the right ticket to the right person so as to maximize the profit has always been a crucial decision for airlines to make during the ticket sales period. In this study, we show how to make the decision regarding the combination of the fare classes to be open, under the premise of a limited number of seats. We propose a two-stage stochastic programming (SP) model, which considers consumers’ choice behavior. In the simulations of the numerical experiment, the effectiveness of the SP model developed in this study are compared with the dynamic programming model (DP) in the literature as well as the first-come-first-served (FCFS) method. In addition to solving the SP model only one time, we have performed the tests to the SP model two and four times. By increasing the number of times for solving the stochastic programming model, i.e., the frequency to adjust the decision for the combination of the open fare classes, the benefits of re-solving is investigated. The experiment results show that by increasing the number of times of re-solving, the revenue is improved. When the number of re-solving reaches four times, the revenue difference when compared with the optimal control from the DP model can be reduced to about 1%. In addition, when the consumer arrival rate increases, the control based on the FCFS method is significantly worse than that of the DP and SP methods |
| Hashtags |
View count:
129