Airline passenger forecasting using neural networks and Box–Jenkins methodology

Demand forecasting for available seats in airlines is important to maximize the expected revenue by setting the appropriate fare levels for those seats. The accuracy of the forecast is the most significant tool of the revenue management systems. The product in airline industry is the seat, which is an expensive, and can’t be stocked. The demand for the seats is almost uncertain, the capacity is constant and difficult to increase and the variable costs are very high. The data used in this study belongs to a major airline company in Turkey and comprise past five years daily passenger data for a flight. It’s used to forecast expected passenger count for that flight for 355 days which are open to sale in reservation systems. Two techniques, Box–Jenkins, artificial neural networks are used for forecasting. Finally, these methods are compared.