Department of Civil and Environmental Engineering, University of North Carolina at Charlotte, Charlotte, North Carolina.
Department of Civil and Environmental Engineering, USDOT Center for Advanced Multimodal Mobility Solutions and Education (CAMMSE), University of North Carolina at Charlotte, Charlotte, North Carolina.
Traffic Inj Prev. 2021;22(7):524-529. doi: 10.1080/15389588.2021.1940983. Epub 2021 Jul 15.
The objective of this research is to identify and compare contributing factors to pedestrian injury severities in pedestrian-vehicle crashes considering both time-of-day and day-of-week.
The pedestrian-vehicle crash data are collected from 2007 to 2018 in North Carolina with categorical factors of pedestrian, driver, vehicle type, crash group, geography, environment, and traffic control characteristics. The final dataset includes 17,904 observations with 69 categorized variables. Four mixed logit models are developed to analyze the crash dataset with segmentations of weekday daytime, weekday nighttime, weekend daytime, and weekend nighttime.
A total number of 31 fixed significant factors and 6 random parameter factors to the pedestrian injury severity are detected in four mixed logit models. According to marginal effects, large vehicle involved, pedestrians with age over 65, hit and run, drunk pedestrian, down/dusk light, dark without roadside light, and industrial land use are identified as the contributing factors that result in more than a 0.08 increase in the probability of fatal injury. Compared to the daytime, most factors are found to have more impact on severe injuries in the nighttime. Also, most factors are found to result in more severe injuries on weekends than on weekdays.
This study identifies and compares the factors to pedestrian injury severity in pedestrian-vehicle crashes considering the temporal variance in time-of-day (i.e., daytime vs. nighttime) and day-of-week (i.e., weekdays vs. weekends). Random effects are explored in mixed logit models. Differences and possible reasons for the significant factors' impact within and across time-of-day and day-of-week are also investigated. Corresponding countermeasures and suggestions to mitigate the impacts of major factors are also discussed, which give practical guidance to planners and engineers, and provide a solid reference to further explore the temporal variance of the crash data.
本研究旨在识别和比较行人和车辆碰撞中行人受伤严重程度的影响因素，同时考虑一天中的时间和一周中的天数。
本研究的数据来源于北卡罗来纳州 2007 年至 2018 年期间的行人-车辆碰撞事故，其中包含行人、驾驶员、车辆类型、碰撞类型、地理位置、环境和交通控制等分类因素。最终数据集包含 17904 个观测值和 69 个分类变量。本研究使用四个混合逻辑回归模型来分析该数据集，其中数据集根据工作日白天、工作日夜间、周末白天和周末夜间进行了分段。
四个混合逻辑回归模型中检测到了 31 个固定显著因素和 6 个随机参数因素对行人受伤严重程度的影响。根据边际效应，大车辆、65 岁以上的行人、肇事逃逸、醉酒行人、黄昏或傍晚光线、黑暗无路边照明以及工业用地等因素被确定为导致致命伤概率增加超过 0.08 的因素。与白天相比，大多数因素在夜间导致更严重的伤害。此外，与工作日相比，大多数因素在周末导致更严重的伤害。
本研究识别和比较了行人和车辆碰撞中行人受伤严重程度的因素，同时考虑了一天中的时间（即白天和夜间）和一周中的天数（即工作日和周末）的时间变化。本研究在混合逻辑回归模型中探索了随机效应。还调查了不同时间和不同时间段内显著因素的影响差异和可能原因。还讨论了相应的缓解主要因素影响的对策和建议，为规划者和工程师提供了实际指导，并为进一步探索碰撞数据的时间变化提供了坚实的参考。
Zotero 插件