城市快速路交通事故持续时间预测

有效的交通事件管理系统需要对交通事故持续时间进行准确的预测。基于北京市122事故报警系统的65000组数据，分别运用多元回归和分类回归树的方法对北京市快速路上的交通事故持续时间进行了建模，并运用另外的8000组数据对预测模型的精度进行了检验，回归模型的误差平均值为30．7463％，分类回归树的误差平均值为29．5197％，误差分析结果表明，多元回归模型和分类回归树模型的结果都比较令人满意；分类回归树并不能显著的提高模型预测的精度。

Incident Duration Models of Urban Expressway

Effective incident management require accurate prediction of incident duration time.In this paper,linear regression and decision tree is employed to model the incident duration time.the 6 5000 incident records from Beijing transportation manage center was used for model development and anther 8 000 for validation.The average error of multiple linear regression models is 30.746 3% while the classification and regression tree model is 29.519 7%.it is shows that the classification and regression tree model can...