Identifying patterns and profiles of rule-violation behaviors among bus drivers: A framework for targeted interventions

Urban bus crashes pose a significant risk to public safety, with driver violations being a major contributing factor. However, the current research focuses primarily on aggregate violation data, overlooking specific patterns associated with distinct categories of violations by bus drivers. This oversight hinders the development of effective interventions tailored to address specific violations. This study aims to bridge the gap by identifying key violations across multiple dimensions and categorizing drivers based on their violation patterns. We proposes a modeling framework that integrates categorical principal component analysis (CATPCA) and K-Means++ clustering algorithms, using observational data from 248 bus drivers with high rates of violation. Violations are grouped into four categories, each corresponding to a distinct dimension of violation behavior. Dimension 1 involves violations related to traffic signs and speed management, dimension 2 involves behaviors of occupying road rights and illegal parking, dimension 3 involves habitual negligence in distracted driving and one-handed driving habits and dimension 4 includes risky behaviors such as running red lights and not yield to pedestrians. Based on these dimensions, drivers were divided into four groups, each of which exhibited differentiated behavioral patterns and psychological tendencies, enabling a deeper understanding of the potential causes of their behaviors. Additionally, this study also includes a detailed analysis of the prominent characteristics of driver groups under different clusters. The research results dispelled the necessity of a “one-size-fits-all” safety strategy and supported the development of targeted intervention measures for bus drivers’ violations, emphasizes the improvement of the quality of internal training.