engleski

Comparison of Air Travel Demand Forecasting Methods

Accurate forecasts of future passenger demand are essential to effective revenue management system. The seat inventory control leans on predictions about the bookings to come to optimally allocate aircraft seats among the various booking classes. Forecasting for airline revenue management systems is inherently difficult because of complex nature of air travel demand which is highly stochastic. The problem is further complicated because of usually great number of origin destination pairs, each with its own seasonal and weekly effects, the economic environment and external factors like competition or special events. The paper describes general problem of forecasting airline demand and compares traditional methods of forecasting (moving averages, exponential smoothing, etc.) against neural networks as a forecasting method. All the methods are compared on the basis of standard error measures.

airline demand forecasting; airline revenue management; time series forecasting; decomposition methods; smoothing methods; neural network forecasting

nije evidentirano