Use of association rules for cause-effects relationships analysis of collision accidents in the Yangtze River

Abstract

In order to discover cause–effect relationships in collision accidents, an association rule-based method is applied to analyze the historical accident data in the Jiangsu section of the Yangtze River from 2012 to 2016. First, the Apriori algorithm is introduced for interesting rules mining, and three types of measures of significance and interest are considered, which are support, confidence and lift. Second, the data are discretized based on previous studies and work experience, and the R software is introduced to facilitate the modeling process. Third, the contributing factors are discovered from the cause-effect relationship analysis. Finally, the generated rules are visualized using the Gephi software to further analysis the unknown relationships and patterns. The observed patterns of collision accidents can be avoided by cutting off some factors in the sequential chain of collision accidents, which is beneficial for prevention of collision accidents. Consequently, this paper provides a data-driven method for accident analysis and prevention.