文章编号： 1672-5328（2026）01-0056-07

李瑞1，杜建坤2
(1. 武汉设计咨询集团有限公司，湖北武汉430014；2. 武汉市规划研究院〔武汉市交通发展战略研究院〕，湖北 武汉430014)

摘要： 与自行车相关的交通事故在各类碰撞事故中占比较高，其诱因通常涉及多因素之间的复杂交 互作用。然而，既有研究多聚焦于单一风险因素对自行车碰撞的影响，对事故诱因间的相互作用机 制尚不明确。为此，提出一种融合潜在类别聚类分析与关联规则挖掘的研究方法，以深入探讨影响 自行车交通事故的风险因素及其交互作用机制。首先，采用潜在类别聚类分析将事故划分为具有明 显不同特征的事故簇；其次，利用关联规则挖掘方法，识别影响自行车交通事故的关键因素及其内 在关联。研究结果表明，社会人口特征、建筑环境、道路网络和土地利用等因素均对自行车交通事 故产生影响，且这些因素在不同事故簇中呈现出显著的异质性效应。

关键词： 自行车交通安全；潜在类别聚类分析；关联规则挖掘；事故特征；建成环境；大伦敦地区

中图分类号： U491.3

文献标识码：A

A Study on the Causes of Bicycle Accidents

Li Rui1, Du Jiankun2
(1. Wuhan Design Consulting Group Co., Ltd., Wuhan Hubei 430014, China; 2. Wuhan Planning & Design Institute [Wuhan Institute of Transportation Development Strategy],Wuhan Hubei 430014, China)

Abstract： Bicycle-related traffic accidents account for a high share of all collision accidents. The causes often involve complex interactions among multiple factors. However, most existing studies focus on the effect of a single risk factor on bicycle collisions, while the interaction mechanism among contributing factors remain unclear. To address this gap, this paper proposes a method that combines latent class clustering analysis and association rule mining, aiming to examine risk factors for bicycle accidents and the interaction mechanism. First, latent class cluster analysis is adopted to group accidents into clusters with clearly different features. Secondly, the major factors and their internal relationships are identified through association rule mining method. The results indicate that sociodemographic characteristics, the built environment, road networks, and land use all affect bicycle accidents. These effects exhibit heterogeneity across different accident clusters.

Keywords： bicycle traffic safety; latent class cluster analysis; association rule mining; accident characteristics; built environment; Greater London