收益管理中考虑顾客策略行为的多航班无约束估计方法

传统的收益管理多航班无约束估计方法假设顾客到达时的购买决策是一次性的,未能充分考虑策略型顾客的跨期替代行为.在仅能获取产品的历史可观察订购量、历史预售开放状态以及市场份额信息的情况下,基于顾客偏好排名列表建立了考虑顾客策略行为的非参数离散选择模型.针对历史预售数据的不完备性,采用EM算法对顾客到达率和非参数离散选择模型的概率质量函数进行联合估计,并提出了考虑历史顾客策略行为的"初始需求"无约束估计计算方法.使用数值算例说明了所提方法的可行性,通过与现有文献中已有方法比较,验证了所提多航班方法能够反映产品价格变化对顾客选择行为的影响,并能更加有效地避免需求预测对未来顾客"初始需求"的高估.

Multi-flight unconstraining method considering strategic customer behavior in revenue management

Traditional multi-flight unconstrianing methods in revenue management assume that customers make one-time purchase decisions at their time of arrival, which can not reflect the inter-temporal substitution behavior of strategic customers. Using only observed historical bookings, product availability data and market share information, a nonparametric discrete choice model considering strategic customer behavior was developed based on the customer rank-based preference lists. For the incompleteness of the historical sales data, the EM algorithm was applied to jointly estimate the arrival rate of customers and the probability mass function of the nonparametric discrete choice model. After that, the unconstraining calculating method for primary demand considering historical strategic customer behavior was proposed. Numerical examples are given to illustrate the feasibility of the proposed method. Through the comparison with other methods in the existing literatures, the results show the proposed multi-flight method can reflect the impact of product price changing on customer choice behavior, and is more effective to prevent overestimating of future customer primary demand in the procedure of demand forecasting.