Predicting maritime accident risk using Automated Machine Learning

In a recent study, published in the journal Reliability Engineering & System Safety, co-authored with Michael André Hindenes Sørli, HyungJu Kim, and Prof. Ilan Alon, we used Automated Machine Learning (AutoML) in analysing maritime accident risk in the Norwegian coastal waters.

Over 40 years of accident records data from the Norwegian Maritime Authority are analysed. A total of 29 ML models are trained using cloud-based AutoML platform DataRobot. The best-performing model is then validated using Python (codes available as supplementary material). The three most impactful factors for accident risk are the category of navigation waters, phase of operation, and gross tonnage of the vessel.

https://doi.org/10.1016/j.ress.2024.110148 (open access)

Video tutorials on the used method are available on RESEARCH HUB YouTube