一种分段在线支持向量回归算法

Segmental online support vector regression algorithm

针对在线支持向量回归算法在提高时间序列预测执行效率的同时,其预测精度会有所下降的问题,提出一种分段在线支持向量回归的时间序列预测方法,通过缩减在线建模数据长度实现快速训练,并对在线支持向量回归模型进行分段存储,根据预测数据与子分段模型的匹配度,选取最优子分段模型预测输出,从而提高在线算法预测精度。通过对黑龙江移动通信话务量时间序列数据的实验结果表明,该算法既很好地保持了在线预测方法的运行效率,又通过分段使预测精度提高了5%～10%。

To improve the poor precision of online support vector regression （Online SVR） when increasing operation efficiency in complicated and nonlinear time series prediction,this paper proposes a novel segmental online SVR algorithm for time series forecasting.Fast training speed is achieved by cutting the training data set short.A segmental strategy is applied and the online SVR model is stored in segments.The most suitable segmental model is selected to output the prediction value according to the matching degree between prediction neighborhood data and all the segmental models.As a result,the forecasting precision is improved.Experiment results with the data provided by China Mobile Communications Corporation Heilongjiang Co.Ltd.show that the proposed method can increase precision by 5% to 10% while keeping the operation efficiency.