基于Multinomial Logit模型的美国北卡罗莱纳州慢行交通事故严重程度分析

Severity Analysis of Slow Traffic Accidents in North Carolina Based on Multinomial Logit Model

为分析慢行(自行车与行人)交通事故严重程度的显著性影响因素,将交通事故分为仅财产损失事故、潜在伤害事故、非伤残事故、伤残事故、死亡事故5类,从人、车、路、环境4个层次选出11个自变量,以交通事故严重程度为因变量,基于Multinomial Logit模型计算每个变量对慢行交通事故严重程度的边际效应。结果表明,驾驶员性别、驾驶员饮酒状态、行人年龄、道路特征、路面状态、一天内发生时间、光线条件、年平均日交通量、地形等9个变量均与道路交通事故严重程度显著相关。

In order to analyze the significant influencing factors of the severity of slow traffic accidents(bicycle and pedestrian),traffic accidents are divided into five categories:property damage only,possible injury,evident injury,disabling injury and fatality.Eleven variables are selected from the four levels of people,vehicles,roads and environment.The traffic accident severity was taken as the dependent variable.Based on the multinomial logit model,the marginal effects of each variable on the severity of slow traffic accidents were calculated.The results show that driver gender,driver alcoholic status,pedestrian age,road characteristics,road surface,time of day,light condition,AADT and terrain are significantly related to the severity of road traffic accidents.