文章编号： 1672-5328（2018）03-0028-05

支野，王大珊，丛浩哲，饶众博
(公安部道路交通安全研究中心，北京100062)

摘要： 现有道路交通事故统计分析技术存在数据项缺失、方法单一、实战应用性弱等问题，很难为 公安交管部门提供针对性的辅助指导。基于深圳市2014—2016 年交通事故数据，采用Apriori 关联 分析算法、贝叶斯理论以及模糊聚类等大数据挖掘方法，探索性地提出道路交通数据缺失数据项填 补、事故伤亡特征因子甄别以及事故危险性分类评价方法。结果表明，该方法可有效提高道路交通 事故数据完整性和事故伤亡特征因子甄别准确性，以及量化交通事故危险度评价。研究方法和结果 可辅助公安交管部门开展道路交通事故预防和交通安全管理工作。

关键词： 道路交通安全；事故统计；Apriori关联分析；贝叶斯；K-means聚类

中图分类号： U491.3

文献标识码：A

Road Traffic Accident Data Analyzing and Its Application: Example of Shenzhen

Zhi Ye,Wang Dashan, Cong Haozhe, Rao Zhongbo
(Road Traffic Safety Research Center of the Ministry of Public Security, Beijing 100062, China)

Abstract： The existing statistical analysis methods for road traffic accidents is problematic because of missing data items, over simplistic, and weak in applications, which make it hard to be useful for traffic management departments. With the Shenzhen accident data from 2014 to 2016, this paper explosively proposes the methods of Apriori algorithm, Bayesian theory and fuzzy clustering big data mining techniques for solving missing accident attribute data problems, identifying accident severity and classifying accident risk. The results show that these methods can effectively improve the accident data integrity, accuracy of characteristic factor selecting for accident casualties, and the assessment of traffic accident risk quantization. The study methods and results can assist traffic management departments in road traffic accident prevention and traffic safety management.

Keywords： road traffic safety; accident statistics; Apriori correlation analysis; Bayesian theory; K-means clustering