Throughput forecasting of different types of cargo in the adriatic seaport Koper. Issue 1 (2nd January 2021)

• Record Type: Journal Article
• Authors: Dragan, Dejan; Keshavarzsaleh, Abolfazl; Intihar, Marko; Popović, Vlado; Kramberger, Tomaž
• Is Part Of: Maritime policy & management. Volume 48:Issue 1(2021)
• Page Start: 19
• Page End: 45
• Publication Date: 2021-01-02
• Subjects: Throughput forecasting -- macroeconomic indicators -- dynamic factor models -- arimax models -- decision support system -- planning in Seaports; Marine resources -- Periodicals; Shipping -- Management -- Periodicals; Policy sciences -- Periodicals; 333.9164
• DOI: 10.1080/03088839.2020.1748242
• Languages: English
• ISSNs: 0308-8839

Abstract

ABSTRACT: An accurate forecasting system has manifested its role as an enabler in supply chains (SC), which makes the operation possible in a maximally synchronized manner. Its applications have gained the attention of scholars across various disciplines such as forecasting in market behavior analysis and tourism industry; material requirement planning in production; transport and logistics foresight in networks and facilities. Seaports, as specific SC members, are not an exception. Accurate forecasting is needed in almost all aspects of the ports' operation to avoid financial losses related to inappropriate investments and planning. The paper addresses the forecasting of joint demand-supply cargo throughputs in the Adriatic Seaport Koper. The research presents a new forecasting approach, namely, DFA-ARIMAX (Dynamic Factor Analysis-ARIMAX modeling). External economic indicators were screened to obtain useful information using the DFA prior to directing the dynamic factors into the ARIMAX forecasting model. The principal component regression and Monte Carlo framework were included to identify indicators that are unique to the port. Findings revealed that a forecasting system by its enriched capabilities to predict the observed throughputs could be seen as Functional Decision Support System. The benchmarking shows that proposed models outperform competitive models. Practical implications are discussed in detail.