Predictive simulation of airline passenger volume based on three models

Abstract

It is of great significance to predict the airline passenger volume accurately no matter for the transport capacity arrangement, the airline adjustment or the planning and development. Considering so many uncertainties and insufficient data in terms of the passenger volume prediction of civil aviation, this paper, based on the daily passenger data of the airline from Beijing to Sanya for the period from 2010 to 2017, applied the random forest prediction model, the support vector regression model and the neural network model to fit the airline data. Upon verification, the mean absolute percentage error (MAPE) of the said three models was 4.18%, 6.87% and 12.38% respectively. In this sense, the random forest prediction model featured the highest prediction precision and the optimal simulation effect in passenger volume prediction.