A comparison of time series methods for forecasting container throughput

Hing Kai Chan, Shuojiang Xu, Xiaoguang Qi

Research output: Journal PublicationArticlepeer-review

40 Citations (Scopus)

Abstract

A good port data management system can undoubtedly improve a port's operations. This, in turn, affects the economic development of the port city or region. Given the fast-changing global environment, forecasting the container throughput of a port is of vital importance. There are many existing time series forecasting methods, which can be used to forecast a port's container throughput. However, there is limited research comparing the performance of different commonly employed methods on the same time series. This study attempts to bridge this gap. This paper first presents several time series forecasting methods, including machine learning-based methods such as Support Vector Regression. Next, these forecasting methods are employed to forecast the port's container throughput using the same set of historical secondary data. Finally, a comparison is made and discussed. Six time series methods were employed for forecasting a port's container throughput and their performances were compared.

Original languageEnglish
Pages (from-to)294-303
Number of pages10
JournalInternational Journal of Logistics Research and Applications
Volume22
Issue number3
DOIs
Publication statusPublished - 4 May 2019

Keywords

  • Port management
  • forecasting
  • machine learning
  • time series

ASJC Scopus subject areas

  • Management Information Systems
  • Business and International Management
  • Strategy and Management
  • Management Science and Operations Research
  • Management of Technology and Innovation

Cite this