Analysis Of Influencing Factors Of Roadside Accident Severity Based On Logit Model

With the development of social economy and science and technology,China’s highway mileage and car ownership continue to grow,transportation infrastructure is becoming increasingly perfect,and road traffic safety problems also increase.In 2020,Traffic accidents in China caused 61,703 deaths and a loss of 1313.61 million yuan,causing huge losses to people’s lives and property.Road side accident is a typical type of accident with serious consequences.Therefore,it is necessary to mine the data of road side accidents,grasp the inherent law of accidents,and provide scientific theoretical support for the development of prevention measures,so as to reduce the occurrence of accidents and reduce the severity of accidents.This paper focuses on exploring the distribution law of roadside accidents and the influencing factors of the severity of roadside accidents.Based on the historical roadside accident data,this paper uses statistical correlation method and discrete model to analyze the data from the perspectives of drivers,vehicles,roads and environment.First of all,combined with the existing research results,the concept of road side accident,road side accident form and road side accident cause are described,and the road side accident is defined in this paper to provide theoretical support for the extraction of road side accident data.At the same time,statistical correlation method and selforganizing Map(SOM)algorithm are used to screen independent variables.Secondly,descriptive statistical analysis is carried out on the distribution law of accident number,casualty number and accident injury degree corresponding to each factor of driver,vehicle,road and environment,so as to grasp the basic distribution characteristics of road accidents.According to the application of discrete model in traffic accident severity,the modeling,solving and checking process of multinomial Logit model and ordered Logit model in accident severity analysis are summarized.Finally,based on the extracted data of 3188 roadside accidents,the influencing factors of roadside accident severity based on Logit model are explored.From the driver,vehicle,road and environment and so on various elements of accident factors,with light,weather,road alignment and vehicle state 16 factors as independent variables,such as accident damage degree as the dependent variable to construct the multinomial Logit and roadside accident severity of ordered Logit analysis model,the backward method is adopted in the process of regression filter significant independent variables,Then the model is solved and tested,the fitting effect of the two models is compared and analyzed,and the influencing factors of roadside accident severity are quantitatively analyzed.According to the research results,reasonable suggestions are put forward to prevent the occurrence of road side accidents and reduce the severity of road side accidents.The results show that compared with the ordered Logit model,the multinomial Logit model can be better applied to the impact analysis of the severity of roadside accidents.The significant variables that affect the severity of road side accidents include 12 factors,such as wrong driving behavior,driver gender,age,ejection,airbag,avoidance behavior,vehicle type,vehicle state,road state,lane number,plane alignment,road speed limit,etc.The confidence interval in the two models is 95% level.Multinomial Logit models also found a 95% confidence interval for sectional alignment.