抄録
This study aims at developing a crash risk estimation model (CREM) considering the interaction of geometry, traffic flow and ambient conditions for basic segments of Nagoya Urban Expressway. A matched case-control study is firstly designed separately for three traffic conditions; low-density, high-density uncongested and congested regimes. Based on these case-control samples, conditional logistic regression is then applied and the CREM is finally developed. The results reveal that the model has statistical significance and acceptable goodness-of-fit, with 86.6%, 80.5% and 70.2% of predictive performance for each of the three traffic conditions, respectively. Regarding crash influencing factors, with the increase of traffic density, the significance of horizontal geometry affecting crash becomes lower. In contrast, the contribution of vertical geometry to crash risk is on the rise. Meanwhile, the effect of average speed on crashes gets more significant. Besides, nighttime or holiday can increase the relative risk compared to daytime or weekday, respectively.
