摘要
针对区域货物周转量的特点:时间上的大波动性和区域之间的密切相关性,指出:反映区域货物周转量预测的指标数据存在高度的非线性、耦合性和时变性。通过对区域公路货物周转量历史数据的分析,确定了相关预测变量。在上述基础上,建立了基于SVR的区域公路货运周转量预测模型。由于该模型建立在统计学习理论的基础上,因而具有良好的非线性数据处理能力,并保证了预测模型的泛化性能。以金华市为应用对象,说明了该模型的有效性。
Aiming at the characteristics of regional freight turnover such as big fluctuation of time,consanguineous relativity among regions,and pointed out the high nonlinearity,coupling and time change character of indexes data which reflect regional freight turnover.Through analyzing history regional road freight turnover data,we confirmed the interrelated forecast variable and established regional road freight turnover forecast model based on support vector regression.Because this model is based on statistical learning theory,it has favorable nonlinear data processing ability,and generalization ability of forecast model also can be ensured.At last,take Jinhua city as an application object to explain validity of the model.
出处
《物流工程与管理》
2011年第3期67-69,95,共4页
Logistics Engineering and Management
关键词
核方法
支持向量回归
公路货物周转量
中期预测
kernel method
support vector regression(SVR)
road freight turnover
medium-term forecast