Data-driven Bayesian network approach to maritime accidents involved by dry bulk carriers in Turkish search and rescue areas

Abstract

In the case of maritime accidents, they cause very serious vital, economic, and environmental problems. As a matter of fact, maritime accidents emerge as an important issue due to the risks brought by the increasing ship tonnage and ship traffic. In this sense, the Turkish Search and Rescue Areas, which include the Aegean Sea islands, the Turkish coasts, and the Turkish Straits, are among the regions with risks in maritime accidents. In this study, the accidents that took place between the years 2001–2019 in the Turkish Search and Rescue Areas were examined, and maritime accident analysis was carried out using Bayesian networks on the accidents involving dry-bulk carriers. Considering accident type as a target variable, this study concentrates on the probabilistic relationships among the factors (i.e., ship type, grt, flag, survival status, navigation region, accident month, accident time, and accident causes) which are thought to influence the occurrence of accidents. According to the research outcomes, it has been determined as probabilistic that there are seasonal and regional differences in the occurrence of accident causes and the occurrence of accidents. In dry bulk carriers, it has been determined that the white flag ships in the Paris Mou ranking are much less likely to be involved in a fatal accident at night. In addition, while the probability of fatal maritime accidents occurring during the daylight hours (06:00–17:59) in accidents in the Çanakkale navigational area is 28%; at night (18:00–05:59), this rate is 41%. Another important matter is that some ships have been found to be involved in accidents repeatedly. In this sense, suggestions have been made for the stakeholders.