Akademik Veri Yönetim Sistemi
International Symposium on Ballast Water and Biofouling Management in IAS Prevention and Control, Antalya, Türkiye, 28 - 30 Kasım 2023, ss.44-45
Port state controls in maritime transportation are the process of inspecting foreign-flagged ships in order to ensure the safety of life and property at sea, prevent marine pollution caused by ships, and improve living and working conditions on ships. The basis of these controls is to ensure that ships comply with international maritime conventions and the standards determined by these conventions. One of these conventions, the “Ballast Water Management Convention”, is a convention that is taken into consideration during port state controls and its compliance is checked by ships. A serious (major) deficiency detected within the scope of this convention may be among the deficiencies that cause the ships to be detained. In this study, an examination was made on the deficiencies arising from the Ballast Water Management Convention in port state controls of container ships, and a model was proposed for the detention risk assessment of ships. A data set was created by examining the reports sharing the results of the controls carried out by the Paris Memorandum (Paris MoU) in the last three years (10.10.2020 – 10.10.2023). Analyses were performed using this data set. Of the total 174 controls subject to the research, 24 resulted in ship detention, and the remaining controls did not result in detention. The number of ships that are detained and those that are not detained in the data set is unevenly distributed. This situation makes the data set an imbalanced data set. In analyses made on unevenly distributed data sets, it is usual for model prediction accuracy to be low. For this reason, balancing unevenly distributed data sets with artificial data is an approach frequently used in the literature. In the methodology applied within the scope of the research, through machine learning algorithms, missing cells were completed, and the data set was balanced. Based on this data set, contributing factors in the detention of container ships, including deficiencies arising from the Ballast Water Management Convention, were determined. A model was generated to predict the risk of ships being detained using machine learning algorithms. It is thought that the research results will contribute to the development processes of inspection preparation strategies and policies of ship management companies and local authorities.