ABSTRACT
Revenue Management (RM) requires the existence of a rigorous demand forecasting mechanism to predict all demand streams on each flight in the air carrier. To define the problem of demand forecasting in RM, consider the hypothetical air carrier network. This air carrier has two major hubs at Denver (DEN) and ORD, where there is one international flight out of ORD to London-Heathrow airport (LHR) and one international flight out of DEN to Frankfurt (FRA). A rigorous demand forecasting enables the correct seat allocation and contributes to air carrier's total revenue. There are several methodologies used for demand forecasting at the itinerary-fare class level including causal modeling, time-series analysis, and adaptive neural network modeling. Exponential smoothing is widely used within the demand forecasting component of the RM system used by major air carriers. Combined forecasting models include the weighted average of historical and advanced booking forecasts, regression methods, and a full information model.
