Non – Stationary Booking Demand in Two-Class Overbooking model

Abstract

Accurate demand forecasting is a crucial component of airline revenue management. However, it is cumbersome to find exact forecasting demand since the historical data, the censor data, do not reflect the actual demand. In order to obtain estimated demand, the Expectation-Maximization (EM) method is used to calculate the estimated demand. Furthermore, the previous research mainly focuses on the stationary demand, which does not represent an actual situation. So, this research engages in finding the optimal overbooking and booking limit of the Two-Class overbooking with non-stationary demand that maximizes the expected profit using numerical experiments. In addition, the performance of the Two-Class overbooking model with non-stationary demand is compared with the booking limit of the airline's policy. The optimal number of update booking limits with non – stationary demand using real data are concerned.

            The simulation finds that the optimal booking calculated mostly is the booking limit. However, the booking limit is suited to the overbooking limit in some situations. Moreover, the expected profit from the optimal overbooking limit is greater than or equal to the airline's booking policy in all study situations. The Two-Class overbooking model yielded up to 21% more the expected profit margins than airline booking policies. In real data study, the Two-Class overbooking model is taken to calculate the optimal number of update booking limits when the demand is non – stationary and it is found that the 16-point update booking limit gives the maximum profit.