Title: Risk analysis of oil tanker accidents in Southeast Asia by using data driven Bayesian network

Authors: Xiaoxing Gong; Yiming Guo; Hanwen Fan

Addresses: College of Transportation Engineering, Dalian Maritime University, Dalian, 116026, China ' College of Transportation Engineering, Dalian Maritime University, Dalian, 116026, China ' College of Transportation Engineering, Dalian Maritime University, Dalian, 116026, China

Abstract: In this study, a new risk analysis framework on oil tanker accidents has been created utilising the benefits of the tree-augmented naïve Bayes (TAN), Bayesian network (BN) and the expectation maximisation algorithm (EM). The framework efficiently tackles the issue of missing data and surpasses the conventional BN models that neglect the connections among the significant variables, thereby improving the accuracy and predictive capability. 25 risk influential factors (RIFs) are identified based on the newly created dataset and related literature. The most high-impact RIFs are identified through sensitivity analysis and risk prediction is performed based on scenario simulation using real case studies to validate models' validity. The study revealed that the top seven RIFs are oil residue ignition, inadequate lookout, dysfunctional management system, impact, gross tonnage, vessel condition, ship age. The model introduced in the paper provides dependable risk prediction results, and will assist to develop critical strategies affecting tanker accidents.

Keywords: maritime safety; maritime accidents; maritime risk; data-driven Bayesian.

DOI: 10.1504/IJSTL.2025.147888

International Journal of Shipping and Transport Logistics, 2025 Vol.21 No.1, pp.39 - 70

Received: 17 Mar 2024
Accepted: 30 Jul 2024
Published online: 07 Aug 2025 *

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