Maritime accident prevention strategy formulation from a human factor perspective using Bayesian Networks and TOPSIS

Human factors contribute to majority of maritime accidents. This study proposes an advanced methodology for maritime accident prevention strategy formulation from a human factor perspective. It is conducted by incorporating Bayesian network (BN) and Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) in a multi-criteria decision-making system. In order to develop rational accident prevention strategies, this work integrates Multiple Correspondence Analysis (MCA), Hierarchical Clustering (HC) and Classification Tree (CT) to generate strategies and describes accident types as criteria for a new multi-criteria risk-based decision-making system. Specifically, MCA is performed to detect patterns of contributory factors explaining maritime accident types. It is complemented by HC and a CT, aiming at creating different classes of vessels. Next, a Bayesian-based TOPSIS model is built to illustrate the features of multiple criteria and the relations among alternatives (i.e. strategies), so as to select the best-fit strategies for accident prevention. The results show that the information, clear order, and safety culture are the three most effective recommendations for maritime accident prevention considering human errors, which presents new insights for accident prevention practice for maritime authorities.