A Comparative Study of Explanatory and Predictive Models in Air Cargo Throughput

Abstract

Air cargo throughput has played an important role in the contribution of total national commerce volume and regional GDP, by the expansion of the global economy. Previous studies have identified the critical significance of air freight in national or international economic growth. Various models, including linear regression and time-series models, have been applied to analyze and predict the factors influencing air cargo volumes and the trends for future development. However, many models are implemented as a single methodology, but a few studies reviewed and mixed explanatory and predictive models at the same time. Therefore, three mainly used methodologies of MLR (multiple linear regression), PCA (principal component analysis), and ARIMA (autoregressive integrated moving average model) are chosen for this paper to compare and analyze the air cargo throughput of one Asia’s representative airport, Shanghai Pudong Airport (PVG). The results of the explanatory model shows that GDP, population, oil consumption, and exchange rate are regarded as the four most important variables influencing PVG air cargo volumes. The outcomes of regressive predictive model presents that MLR-PCA has a better performance than MLR, while in ARIMA predictors, the ARIMA (p:12, d:0, q:0) is shown to have a superior predictive fit than the ARIMA (p:1, d:0, q:0). By comparing different methodologies, this paper contributes to the industry’s study of the variables affecting air cargo and the future improvement of air cargo throughput forecasting capabilities.