Integrating behavioral theory and ANNs for understanding electric bikers’ red-light running behavior

Abstract

Understanding red-light running (RLR) behavior among electric bikers (e-bikers) is critical for addressing the high accident rates associated with this behavior. Traditional analytical methods, such as econometric modeling, often fail to capture the non-linear dynamics of traffic violations, limiting their effectiveness in exploring the complexity of such behaviors. Conversely, Artificial Neural Networks (ANNs) excel in handling non-linear relationships but lack interpretability, making their application in decision-making challenging. This study introduces an innovative six-step analytical framework that integrates hybrid ANNs with the Theory of Planned Behavior (TPB). This integration utilizes a network weight-based approach to quantify the impacts of influencing factors within the ANNs. The results demonstrate that this hybrid framework not only enhances predictive accuracy but also provides a deeper understanding of the motivational drivers behind e-bikers’ RLR behavior. The study identifies significant behavioral heterogeneities across e-biker groups, emphasizing the need for targeted interventions. Based on these findings, a multi-faceted intervention strategy is proposed, combining educational campaigns, regulatory measures, and community engagement efforts tailored to distinct behavioral profiles. This research provides a robust foundation for developing safety improvement programs that aim to reduce e-biker accidents and improve overall road safety.