A data-driven method for modeling human factors in maritime accidents by integrating DEMATEL and FCM based on HFACS: A case of ship collisions

A data-driven method is proposed to measure the static and dynamic interrelations between human factors contributing to maritime accidents. The proposed method integrates the superiorities of the HFACS (Human Factors Analysis and Classification System), DEMATEL (Decision-Making Trial and Evaluation Laboratory), and FCM (Fuzzy Cognitive Map). First, a maritime accident scenario is defined based on 240 ship collision accident reports, then the human factors at different levels are identified and structured within the HFACS framework under the guidance of grounded theory. Second, the DEMATEL method is adopted to determine the causal relationship between human factors and the importance ranking of each factor based on historical accident statistics. The FCM model is developed in the end to realize dynamic prediction and diagnostic inference of human factors involved in maritime accidents. More objective and realistic results are presented by applying the proposed method without the necessity of expert judgment. The results show that Lookout negligence, Poor safety management practice of the shipping company, and Failure to take effective collision avoidance actions promptly are the top three important human factors, additionally, Not familiar with COLREG has the highest influencing degree and Lookout negligence has the highest influenced degree in the human factors network.

Automated summary

A data-driven method is proposed to measure the interrelations between human factors contributing to maritime accidents. The method integrates HFACS, DEMATEL, and FCM to analyze ship collision accident reports, identify human factors, and determine their causal relationships and importance ranking. The results provide objective and realistic insights into the important human factors and their impact in maritime accidents.