Citation

Wang L, Huang R, Shi W, Zhang C. Transp. Policy 2021; 103: 236-244.

Copyright

(Copyright © 2021, Elsevier Publishing)

DOI

10.1016/j.tranpol.2021.02.006

PMID

unavailable

Abstract

Marine accidents cause not only significant economic losses, but also severe environmental pollution and inestimable human casualties, which have become a worldwide concern. To better cope with this concern, this paper adopts temporal association rules (TARs) to mine and discover the domino effect in marine accidents. Using the dataset of 5754 marine domino accidents (MDAs) collected from the International Maritime Organization and IHS Markit Company, the main findings of this paper are as follows. First, 'hull damage' was found to be the most frequent accident in MDAs, and 'collision' was more likely to cause the damage in the whole hull. Second, 'oil spill' was most often observed as a final marine accident. Meanwhile, 'foundered' was more likely to cause 'oil spill' in both oil tanker and general cargo ship MDAs. Third, it is pointed out that most probable scenarios involved 'hull damage' as the basic accident which ended with 'foundered' and 'oil spill' as top accidents. These findings not only advance our knowledge of marine accidents from the perspective of the domino effect, but also provide insights into improving marine safety.

Language: en

Keywords

Domino effect; Machine learning; Marine accidents; Risk factors; Temporal association rules