Research on the risk influential factors of container ship accidents based on association rules mining and complex network

Abstract

This study combines association rule mining and complex network theory to explore the key risk influential factors (RIFs) of container ship accidents and their interrelationship. Firstly, based on the 103 container ship accident investigation reports, five categories of accident RIFs such as human factors, ship factors, cargo factors, environmental factors and management factors are identified. Second, the Apriori algorithm is applied to find out the correlation between accident RIFs and the correlation between RIFs and each type of accident. Third, complex network theory is then applied to establish a vector-weighted network of container ship accident RIFs, and comprehensive network visualization is subsequently conducted. Finally, topological feature analysis is applied to comprehensively examine associations between RIFs, and robustness analysis is used to find the key container ship accident RIFs. The analysis reveals that collisions and groundings are the most frequent accidents in container ship accidents, with human factors (e.g. negligent lookout) and management factors (e.g. improper bridge resource management) being the primary RIFs. The interaction between human and management factors is most significant. Network topology analysis highlighted high-degree RIFs, enabling targeted risk mitigation strategies. This study aids in disrupting accident RIFs networks and supports intelligent shipping safety mechanisms and risk management optimization for maritime companies.