Driving risk assessment using near-crash database through data mining of tree-based model

•This paper reports a comprehensive naturalistic driving experiment to collect driving data under potential threats on actual Chinese roads.•A novel method is proposed to quantify the driving-risk in a near-crash by clustering the braking characteristics.•This paper explores the factors that greatly influence the driving-risk level by employing CART.•The results indicate that velocity when braking, triggering factors, potential object type and potential crash type exerted the greatest influence on the driving-risk levels. This paper considers a comprehensive naturalistic driving experiment to collect driving data under potential threats on actual Chinese roads. Using acquired real-world naturalistic driving data, a near-crash database is built, which contains vehicle status, potential crash objects, driving environment and road types, weather condition, and driver information and actions. The aims of this study are summarized into two aspects: (1) to cluster different driving-risk levels involved in near-crashes, and (2) to unveil the factors that greatly influence the driving-risk level. A novel method to quantify the driving-risk level of a near-crash scenario is proposed by clustering the braking process characteristics, namely maximum deceleration, average deceleration, and percentage reduction in vehicle kinetic energy. A classification and regression tree (CART) is employed to unveil the relationship among driving risk, driver/vehicle characteristics, and road environment. The results indicate that the velocity when braking, triggering factors, potential object type, and potential crash type exerted the greatest influence on the driving-risk levels in near-crashes.