Analysis Of Influencing Factors Of Fatal Traffic Accidents Based On Defensive Driving Behavior

This paper starts from the perspective of defensive driving behavior and explores how relevant independent variables affect the driver's defensive driving behavior.Four key contents of this research were identified: defensive driving behavior,the combination of parametric and non-parametric models,semiparametric additive regression model,and interaction between independent variables.It aims to analyze whether the driver can find potential risk factors in time and take effective avoidance measures before the road traffic accident occurs,the avoidance of reducing the severity of the traffic and reduce the casualties and property losses caused by traffic accidents.At present,although defensive driving technology has been applied in many fields,there are few studies on defensive driving behavior at home and abroad,and even fewer studies on defensive driving behavior based on traffic accident data.This paper takes the 2017 Fatal Accident Reporting System(FARS)traffic accident data compiled and recorded by the National Highway Traffic Safety Administration(NHTSA)as the research content,through reading a lot of relevant literature,the research direction and related variables of this article are determined,whether the driver takes avoidance measures as a characteristic variable of defensive driving behavior.Using various statistical models,the relationship between the driver's defensive driving behavior and eight independent variables such as age,gender,and travel speed in a fatal traffic accident is analyzed,and the influencing mechanism between important influencing variables and whether drivers take avoidance measures was analyzed in detail,including independent effect analysis of independent variables and interaction effect analysis of independent variables.First,the binary logistic regression parameter model and the CART decision tree non-parametric model are used to rank the influence of the eight independent variables and the dependent variable(whether the driver manually takes avoidance measures before the accident occurs),and the results of the two models are integrated.It is concluded that the most important independent variable affecting the dependent variable is the travel speed,followed by the two independent variables of the accident happened time and age.In the semiparametric additive regression model,using individual effect analysis and interactive effect analysis,using travel speed as the main variable,the accident happened time and age as the covariate,the internal relationship between vehicle travel speed and whether the driver takes avoidance measures is analyzed.Substitute covariates into the model to investigate whether there is a difference in the relationship between travel speed and dependent variable when there is an interaction between independent variables.The results of the study show that,in the analysis of individual effects,when the travel speed of the driver is at different stages,there is a large difference in the possibility of taking avoidance measures.When the travel speed is less than 60km/h,the increase of the speed stage of the driver will greatly increase the possibility of the driver taking avoidance measures;when the travel speed is less than 140km/h,the possibility of the driver taking avoidance measures will continue to increase in a state of decreasing speed;when the speed exceeds the equilibrium point of 140km/h,there is not enough reaction time to make avoidance measures even with high vigilance.In the analysis of interaction effects,because of the addition of covariates,lead to change of the curve between speed and the dependent variable,which also suggests that the interaction between independent variables will affect the results,the results show that when maintaining the same travel speed,the driver's sense of defensive driving is relatively weak at three levels: driving at night,young drivers and old drivers.