Forecasting and optimization of fare families

In this article we study the general forecasting and optimization problem for fare families. Forecasting is based on a fare family choice model, which explicitly represents passenger travel purpose, willingness to pay and product preference. By applying the fare adjustment theory developed by Fiig et al (2010) it is possible to map the fare family choice model to an equivalent independent demand model that can be used to calculate optimal policies using standard optimization algorithms. The optimal policies derived from the solution may not be nested by fare order (higher valued flexible products are closed while a lower valued restricted product is open). A simulation study in a competitive revenue management network using the Passenger Origin and Destination Simulator shows substantial performance gain in both revenue and load factor compared to hybrid forecasting.