Seafarer competency analysis: Data-driven model in restricted waters using Bayesian networks

Despite the efforts of maritime authorities to enhance seafarer competencies through the International Convention on Standards of Training, Certification and Watchkeeping for Seafarers (STCW), human error remains a leading cause of maritime accidents. To thoroughly investigate the impact of various human errors among seafarers on accidents, this paper aims to examine the relationships between seafarer competencies and maritime accidents using a data-driven approach from the perspective of bridge resource management (BRM). Through analysis of historical maritime accident reports, the dataset of seafarer competencies associated with maritime accidents is established. The least absolute shrinkage and selection operator (LASSO) method is employed to identify the critical seafarer competencies for accident prevention. Then, a data-driven Bayesian Network (BN) model, based on a Tree Augmented Naive Bayes (TAN) method, is constructed to reveal the relationship between seafarer competencies and accident types, which are validated by sensitivity analysis and case study. The results indicate that the key seafarer competencies for all maritime accidents are 'Maneuvers', 'Amend/maintain ship course', 'Decision making', 'Cognitive capacity', 'Information', 'Procedure operations', 'Situational awareness' and 'Communication'. Moreover, the study underscores the importance of leveraging lessons learned from past accidents to mitigate risks and ensure safe maritime operations. The findings contribute to a deeper understanding of the dynamics between seafarer competencies and accident types, unveiling the joint impact of different seafarer competencies on maritime accidents. This perspective offers valuable insights for maritime authorities in strengthening maritime safety regulations.