Study on composite forecasting model of air passenger capacity based on air partition

Abstract

Firstly, an air passenger capacity investigation at the capital international airport is made, and a composite forecasting model based on total air passenger capacity is established, in which multiple regression and ARIMA model are parallel connection and their forecast results are series connection with BP neural network. Secondly, according to the average growth rate of air passenger capacity, all airlines are divided into 5 subareas, and the series connection model of ARIMA and BP neural network is established. Finally, short-term air passenger capacity at the capital international airport is forecasted by the composite models, and analyzed results show that the model based on air partition is more precise than the model based on total air passenger capacity, which is a kind of viable and practicable air passenger forecasting model.