Based on the driver’s driving habits and behavior in driving, the driver’s driving style can be divided into aggressive or normal, driving style has a significant impact on driving safety, road traffic efficiency and vehicle energy consumption and so on. Accurate driving style evaluation is essential to improve driving safety and reduce energy consumption. This study proposes a driver driving style evaluation model based on lane change behavior using clustering algorithm. The study extracted 2,861 lane change segments of 16 drivers from naturalistic driving data and then these lane change behavior were analyzed under the “comparable environment” in which the external traffic environment (including road facility type, traffic congestion, weather conditions, etc.) is basically the same. Here, ”comparable environment” is used to maximize the elimination of the traffic environment interference to the driver’s driving behavior. This study also discusses the sample size required for feature extraction. The research shows that different types of drivers have significant differences in the duration of the lane change, the forward acceleration, the lateral acceleration, and the TTC. Based on these lane change behavior characteristics, we used the clustering algorithm to classify the driver’s driving style. The assessment found that the driver who was evaluated as aggressive not only showed greater forward acceleration and shorter duration in the lane change behavior but also was more serious in speeding behavior than other types of drivers which shows consistency in driving habits.

CICTP