基于聚类分析的交通事故影响因素研究

为研究道路交通事故的主要因素,先对数据变量集进行赋值,使用Calinski-Harabasz函数选择适宜聚类簇数,利用K-means聚类算法聚类交通事故数据,再采用Logistic模型对事故数据建立严重程度分类模型。结果表明,基于Calinski-Harabasz函数的K-means聚类的Logistic回归,既考虑k值选择的合理性以及数据的异质性,又提高了模型的精度;星期、路面状况、地形、路口路段类型仅在某一个类别中显著。事故发生时间、年龄、交通方式、驾龄、季节、人员类型和能见度在两个及以上类别中显著。能见度、人员类型在多个类别中显著,但是影响方式不同。可知,上述变量对交通事故的严重程度影响具有差异性。

In order to study the main factors of road traffic accidents,the data variable set is first assigned,the Calinski-Harabasz function is used to select the appropriate number of clusters,the K-means clustering algorithm is used to cluster the traffic accident data,and then the Logistic model is used to establish a severity classification model for the accident data.The results show that the Logistic regression based on K-means clustering based on the Calinski-Harabasz function not only considers the rationality of k value selection and data heterogeneity,but also improves the accuracy of the model;week,road condition,terrain,and intersection section type are only significant in a certain category.Accident time,age,transportation mode,driving age,season,personnel type,and visibility are significant in two or more categories.Visibility and person types are significant in multiple categories,but the impact is different.It can be seen that the above variables have different impacts on the severity of traffic accidents.