The Study On Pedestrian Crashes Analysis Based On Two Level Logit Model

Pedestrian crash has received more and more attention in recently years.As the most vulnerable group on urban roads,the importance of pedestrian safety research is obvious.It is significance to study the rules and influencing factors of pedestrian fatal accidents and put forward targeted preventive measures,which can reduce the occurrence of pedestrian traffic crashes,reduce the severity of pedestrian traffic crashes and improve the safety on urban roads.Based on existing studies,most of the pedestrian crashes analysis scenes of urban roads are based on the analysis of single scenes of urban roads,most of those are intersections of urban roads.In fact,urban road is a spatial whole system,the factors which have effect on urban road intersections and urban road sections may be different from each other due to different spatial distribution,vehicle attributes and pedestrian characteristics.This paper uses the data of the metropolitan area in Denver,Colorado from 2006 to 2016 and classifies the overall data into three categories:urban road intersection,urban road section and driveway access.The model which was built in this paper is a two-level model.The first level model is a Logit model for pedestrian accident severity prediction.The second level model considers the potential influencing factors between metropolitan cities,and is expressed by setting a random intercept term in the model.This paper proposes the thought of using bayesian probability inference to calculate the parameters rather than the maximum likelihood method,setting parameter distribution for each parameter,using Monte Carlo Markov(MCMC)algorithm to generate model parameter distribution,realize the visualization of model parameter calibration,and then infer parameter values.At the same time,this paper aims to do comparison of the traditional model prediction results with the model prediction results which is according to the bayesian probability inference.The results show that the model prediction accuracy of bayesian probability inference is higher than that of the model prediction of urban road intersection and urban road section.According to distribution regularity of summary in this paper,the urban road pedestrian traffic accident,analysis of factors affecting pedestrian traffic accident,from the people,vehicles,roads,environment four aspects proposed 16 main influenced factors.Modeling prediction analysis results show that:(1)the road conditions,the weather conditions,the pedestrian gender,the driver's speed and the accident casualties five factors caused urban road intersection pedestrian casualties significantly influence.Once a pedestrian traffic accident occurs in snowy day,the pedestrian death or injury probability is 62.08%;The probability of pedestrian traffic accidents with dry pavement causing pedestrian casualties is 56.97%.(2)Six factors which are namely,the lighting condition,the weather condition,pedestrian age,the driving direction,whether the driver uses seat belt and whether the accident happens on weekdays significant influencing factors for pedestrian casualty crashes on urban road sections.In pedestrian traffic accidents on sunny roads,85.96%of the time,pedestrians are killed or injured.Under the light condition of early morning and dusk,the probability of pedestrian traffic accident resulting in pedestrian death or injury is as high as 82.62%.(3)The vehicle type is the significant factor that causes pedestrian fatal crashes at the entrance of the driveway access.The increase of SUV vehicle type makes the probability of pedestrian fatal crash at the driveway access.