最小二乘支持向量机的扩展及其在时间序列预测中的应用

Generalization and application in time series forecasting of the least square support vector machine method

根据时间序列近期数据较远期数据包含有更多未来信息的思想,对最小二乘支持向量机预测方法进行了扩展,得到了更具一般性的最小二乘支持向量机预测模型,给出了扩展后的预测模型具体算法。两个时间序列的预测实例表明,扩展后的预测方法获得了更好的预测效果,提升了最小二乘支持向量机预测方法的价值。

According to the theory that the present data contains more future information than historical data in time - series, the paper extends the prediction method of least square support vector machine and obtains a more general prediction model of least square support vector machine, and develops algorithm of the extended prediction model. Prediction examples of two time - series show that the extended model is more effective. Therefore it improves the value of the prediction method of least square support vector machine.