Title: A systematical forecasting method for container throughput of correlated ports: a case study of Shenzhen port and Hong Kong port

Authors: Lulu Zou; Guowei Hua

Addresses: Panyapiwat Institute of Management, 85/1 Moo 2 Chaengwattana, Amphoe, Pak Kret, Nonthaburi 11120, Thailand ' Panyapiwat Institute of Management, 85/1 Moo 2 Chaengwattana, Amphoe, Pak Kret, Nonthaburi 11120, Thailand

Abstract: Current studies on container throughput forecasting are mainly focused on independent forecasts of individual ports, neglecting the deep underlying correlation between the ports and thus may lead to large errors of the prediction. To overcome the weaknesses, this paper proposes a new container throughput forecasting method to systematically forecast the correlated ports. A systematical forecasting model (SFM) is established based on the correlation between the ports identified by the Granger causal test and estimated using the method newly proposed in this paper. For verification purposes, multiple forecasting models, including the newly proposed SFM and the independently forecasting models, are constructed and compared in terms of the forecasting performance based on the monthly container throughput data of Shenzhen port and Hong Kong port, the empirical results show that the new model is superior to its rivals in terms of absolute prediction accuracy and direction accuracy.

Keywords: container throughput forecast; correlated ports; Granger causal test; ANN; artificial neural network.

DOI: 10.1504/IJSOI.2018.097488

International Journal of Services Operations and Informatics, 2018 Vol.9 No.4, pp.297 - 311

Received: 19 Mar 2018
Accepted: 24 Mar 2018
Published online: 24 Jan 2019 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article