Impact of external factors on air passenger demand prediction using machine learning

Abstract

Do external factors impact the volume of air passengers? Can they be employed to analyze the dependency of hidden parameters in demand forecast strategies? In this research, a framework is proposed to investigate the impact of external factors on demand for air travel by combining the features extracted from various platforms with the historical volume of inbound passenger data from January 2011 to December 2021 and comparing the information symmetry to uncover relations between the data, proving whether they contributed to the shift in demand. A selection of machine learning regression models, namely, gradient boosting, random forest, and support vector regression, were utilized to build a prediction model with and without the inclusion of the additional variables. Their performance will justify our assumption of the impact of external factors on passengers traveling by air. Employing Thailand’s historical inbound passenger volume, the result had shown that with the addition of explanatory variables had reduced RMSE. A combination of certain weather elements and search queries has the most impact on the air travel demand in Thailand, but the combination varies in each region. Event indicators and econometric variables introduce further enhancement in accordance with the preliminary assumption of their influences on the volume of passengers.