Key points are not available for this paper at this time.
Identification of risky driving behaviours throughout a driver's journey and timely implementation of warning and corrective measures are crucial for proactive prevention of traffic accident risk. By identifying and evaluating risky driving behaviours during the journey, driving trajectory data serves as a foundation for proactive traffic accident prevention. This study involves the collection of driving trajectory data from drivers of different types of vehicles via onboard On-Board Diagnostics (OBD) devices. Based on journey segmentation, driving behaviour indicators were calculated. Then the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS)-Principal Component Analysis (PCA) method was utilised to form the basis for driving risk assessment, and machine learning techniques were employed to predict driving risks, while the accumulated local effects (ALE) methodology was implemented to examine the interactions among the influence factors. The results indicate that the interplay between jerk, acceleration/deceleration, and other driving behaviours adheres to a complex nonlinear pattern, exerting multifaceted influences on driving risk. This study advances the understanding of driving behaviours and their implications for traffic safety, providing a foundation for the development of more holistic and impactful traffic safety strategies.
Guo et al. (Thu,) studied this question.