Applications of machine learning methods in port operations – A systematic literature review

• Published in: Transportation research. Part E, Logistics and transportation review Vol. 161; p. 102722
• Main Authors: Filom, Siyavash, Amiri, Amir M., Razavi, Saiedeh
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.05.2022
• Subjects: Container terminals; Data analytics; Machine learning; Port; Seaport; Systematic literature review; Seaport; Data analytics; Container terminals; Port; Machine learning; Systematic literature review
• ISSN: 1366-5545; 1878-5794
• DOI: 10.1016/j.tre.2022.102722

•A structured literature review on application of machine learning methods in port operations is performed.•70 relevant articles are divided using four different categorizations (Operation, method, type, and data).•Evidence from the literature present that ML-based methods exhibit great potential in increasing port performance.•The conducted review reveals that ML methods are increasingly contributing to decision-making procedures besides their conventional predictive role.•There exist two considerable gaps in the literature: (1) most of the available data remained underutilized; and (2) there is lack of real-time use cases of ML applications in port operations. Ports are pivotal nodes in supply chain and transportation networks, in which most of the existing data remain underutilized. Machine learning methods are versatile tools to utilize and harness the hidden power of the data. Considering ever-growing adoption of machine learning as a data-driven decision-making tool, the port industry is far behind other modes of transportation in this transition. To fill the gap, we aimed to provide a comprehensive systematic literature review on this topic to analyze the previous research from different perspectives such as area of the application, type of application, machine learning method, data, and location of the study. Results showed that the number of articles in the field has been increasing annually, and the most prevalent use case of machine learning methods is to predict different port characteristics. However, there are emerging prescriptive and autonomous use cases of machine learning methods in the literature. Furthermore, research gaps and challenges are identified, and future research directions have been discussed from method-centric and application-centric points of view.