Research On The Prediction Model Of Traffic Accident Severity Of Two-lane Highway Based On Logit Model

China's mileage of highway mileage has grown rapidly,the number of car ownership has increased year by year.The colleagues who brought the convenience of transportation to our people also led to the occurrence of traffic accidents.In recent years,the traffic safety of twoLane highways has improved,but there are still relatively high accidental injury rates and death rates.To conduct research on traffic safety,the focus is not only on avoiding traffic accidents,but also on reducing the severity of traffic accidents.To minimize the severity of traffic accidents is another way in the event that the traffic accidents are not be prevented.This thesis takes the two-Lane highway traffic accident as the research object,starting with the four factors of the driver,the vehicle,the road and the environment,and analyzing the potential factors affecting the severity of the traffic accident on the two-Lane highway.Based on the actual traffic accident data of 321 two-Lane highways in an area in China,20 independent variables were selected and the quantitative method of independent variables was explained in detail.Then,on the basis of summarizing the existing research results,a model for predicting the severity of traffic accidents on two-Lane highways is constructed,and the factors that significantly affect the severity of traffic accidents on two-Lane highways are analyzed.Calibration results for ordered response models,includes the ordered response model ordered Logit model and partial proportional odds model,we got that,whether the perpetrators drink alcohol,the number of vehicles involved in the accident,whether it is related to motorcycles,whether it is related to pedestrians,whether there are protective facilities on the roadside,whether it is a holiday,whether it passes through residential areas significantly affected the severity of the traffic accident on two-Lane highway traffic accidents.By testing the goodness of fit and the accuracy of the prediction of the ordered response model,it is found that the partial proportional odds model is superior to the ordered Logit model,and partial proportional odds model are more suitable for modeling the severity of two-Lane road traffic accidents than the ordered Logit model;Through the analysis of the results of the disordered response model,includes multinomial Logit model and mixed Logit model,whether the perpetrators drink alcohol,the number of vehicles involved in the accident,whether it is related to motorcycles,whether it is related to pedestrians,whether there are protective facilities on the roadside,whether it is a holiday,whether it passes through residential areas have a significant impact on the severity of a twotwo-Lane highway traffic accidents,through the test of goodness of fit and prediction accuracy of the disorder response model,the mixed Logit model is superior to multiple Logit models.The mixed Logit model is more suitable for modeling the severity of two-Lane road traffic accidents.According to the analysis of the factors affecting the severity of two-Lane road traffic accidents,especially according to the significant influence factors obtained from the model,we can prevent two-lane highway traffic accidents and provide decision-making reference to improve the two-Lane traffic conditions.