美国萨克拉门托地区2022年1-8月份高速公路交通事故数据集  

A dataset of freeway traffic accidents in the Sacramento Area,USA from January to August 2022

作　　者：冯健[1] 李珂 余浪 刘天 牟琦[1] FENG Jian;LI Ke;YU Lang;LIU Tian;MU Qi(School of Computer Science and Technology,Xi'an University of Science and Technology,710600,China)

机构地区：[1]西安科技大学计算机科学与技术学院,西安710600

出　　处：《中国科学数据(中英文网络版)》2025年第1期440-452,共13页China Scientific Data

基　　金：国家重点研发计划(2022YFB3304401)。

摘　　要：本研究构建了一个涵盖多源数据的美国萨克拉门托地区2022年1-8月高速公路交通事故数据集,旨在解决现有交通事故数据集稀缺且数据类型单一的问题。数据集整合了交通流量、检测器位置、交通事故详情以及天气状况等多维度信息,通过数据采集、预处理和匹配等步骤,共包含59,626,368条记录。数据集采集自美国加利福尼亚州交通运输部PeMS系统和美国国家气象局,涵盖萨克拉门托地区6条主要高速公路。数据预处理包括特征字段选择、重复值去除、缺失值填充和数值编码,确保数据的一致性和完整性。数据匹配则通过时间匹配、检测器匹配和特征融合,将事故记录与最近的检测器关联,并获取流量和天气信息。数据集具有标准化的数据格式和高质量的数据内容,适用性广,支持交通事故预测、交通流量预测和路网结构研究等多种研究方向。其构建方法可为其他地区类似数据集的开发提供参考框架,同时为国内研究人员在数据公开性和多样性不足的情况下开展本地化研究提供支持。This study presents a comprehensive multi-source traffic accident dataset for the Sacramento area freeway network from January to August 2022,addressing the scarcity and limited data diversity in existing traffic accident datasets.The dataset integrates multi-dimensional information,including traffic flow dynamics,detector locations,accident records,and weather conditions,totaling 59,626,368 entries collected from the California Department of Transportation’s Performance Measurement System(PeMS)and the National Weather Service.Covering six major freeways in the Sacramento area,the dataset was constructed through systematic data collection,preprocessing,and matching procedures.The preprocessing steps encompassed feature selection,duplicate removal,missing value imputation,and numerical encoding to ensure data consistency and integrity.A three-phase matching methodology(temporal alignment,detector proximity mapping,and feature fusion)was employed to effectively associate accident records with corresponding traffic detectors and meteorological data.With a standardized format and high data quality,this dataset supports a wide range of research applications,including traffic accident prediction,flow pattern analysis,and road network optimization.Additionally,its methodological framework provides a replicable model for researchers working in data-scarce environments,enabling localized studies where public data is limited.The open architecture of this multi-dimensional dataset facilitates comparative analyses and enhances opportunities for data-driven transportation safety research.

关 键 词：萨克拉门托地区 交通事故 交通流量 天气 

分 类 号：U491.31[交通运输工程—交通运输规划与管理]