| The rapid increase of car ownership in China has brought convenience for residents to travel,but also brought many traffic safety problems.The Sichuan-Guizhou region is located in the southwest China with complex geographical conditions and low level of urbanization.In China,the researches on traffic safety mainly focus on urban roads in the eastern and central regions.However,studies on western regions,especially non-urban roads,are scarce.The west is also the place where traffic accidents frequently occur,and traffic accidents have always threatened the safety of life and property of residents.The traffic safety problem in the western region needs to be solved urgently.This paper starts with the China In-depth accident study database,combined with Latent Class Analysis(LCA)and Ordered Probit(OP)model to analyze the influence of road environment and vehicle characteristics on the accident severity in typical regions of Sichuan and Guizhou.LCA divides all cases into 4 classes:(1)two-vehicle accidents involving trucks;(2)pedestrian accidents and multi-vehicle accidents;(3)two-wheeler accidents;and(4)single-vehicle accidents.The OP model requires a smaller sample size than other analytical models,which is suitable for this study.Driver characteristics are the most important characteristics of accident causes.In this paper,Frequent Pattern Growth(FP-Growth)association rules algorithm is used to explore the relationship between driver-related characteristics and accident severity,and 100 strong association rules are obtained.The main results of this paper are as follows:(1)This paper explored the traffic conditions,the distribution of persons,vehicles,road environment and other characteristics in the accident data,and the proportion of accident severity in each characteristic variable on non-urban road in typical regions of Sichuan and Guizhou.These provide the condition for analyzing the severity of accident.(2)This paper explored the variables that have a significant impact on the severity of the accident in the characteristics of the road environment and vehicles in the typical regions of Sichuan and Guizhou and quantified their impact degrees.The results illustrate that the characteristics of these regions are unique.Truck accidents account for 46.21%proportion of cases,where cases involving trucks tend to significantly increase accident severity,except for single-vehicle accidents(marginal effects of fatal injuries:+0.219,+0.053,and+0.2).Road environment characteristics such as rain,straight path and general bend and slope road could increase single accident severity(+0.294,+0.122,+0.157).These regions also share many similar characteristics with other regions,like travel at night(class 1,2,4;+0.219,+0.053,+0.2),travel in summer(class 1,3;+0.13,+0.206)and winter(class 1,3;+0.23,+0.225)could increase accident severity.Compared with day travel,travel at night is more likely to lead to serious injury in classes 1,2,and 4.In classes 1 and 3,summer and winter(+0.23,+0.225)could increase accident severity compared with spring.In addition,this study found that variables like road alignment,road cross-section profile,vehicle lights,and trucks also had a significant impact on accident severity.(3)This paper explores the association rules between driver-related characteristics and accident severity on non-urban road in typical regions of Sichuan and Guizhou.7 of the death association rules involved pedestrians,with the highest confidence of 94.3%.There were also more than 5 association rules for 1.5-2.5 hours and death after non-intersection.In addition,gender,age stage,and road familiarity also have strong association rules with accident severity.This research can put forward some targeted suggestions for regional transportation departments,auto OEMs and parts manufacturers,local residents and tourists,so as to improve the level of road traffic safety in Sichuan and Guizhou.In addition,findings of this paper can provide support for regional accident characteristics and traffic research in China. |
