New Formulations for Price and Ticket Availability Decisions in Choice-based Network Revenue Management

摘要

We consider the network revenue management (RM) problem under a MultiNomial Logit (MNL) discrete choice model. We present two mathematical formulations that build on previous research, focusing on reducing the size and complexity of the problem while maintaining an accurate representation of passenger choice and industry practices. The first formulation addresses the computational issues of traditional network revenue management by introducing a Mixed Integer Program (MIP) that reduces the number of variables needed to consider while maintaining revenue performance versus other models. The second formulation eliminates the need for static pricing of fare classes, as seen in most RM literature, and considers a continuous pricing decision linked with ticket availability, directly incorporating passenger sensitivities to price and other attributes. Both models perform well in example test cases, improving on popular leg- and network-based RM methods used in practice.