基于ARIMA和LSSVM的非线性集成预测模型

针对复杂时间序列预测困难的问题,在综合考虑线性与非线性复合特征的基础上,提出一种基于ARIMA和最小二乘支持向量机(LSSVM)的非线性集成预测方法.首先采用ARIMA模型进行时间序列线性趋势建模,并为LSSVM建模确定输入阶数;接着根据确定的输入阶数进行时间序列样本重构,采用LSSVM模型进行时间序列非线性特征建模;最后采用基于LSSVM的非线性集成技术形成一个综合的预测结果.将该方法用于中国GDP预测取得的结果,与单独预测方法及流行的其他集成预测方法相比,预测精度有了较大的提高,从而验证了方法的有效性和可行性.

A Novel Nonlinear Ensemble Forecasting Model Incorporating ARIMA and LSSVM

In order to solve the problem of complex time series forecasting including the linear and nonlinear features,a new ensemble forecasting model incorporating ARIMA and LSSVM is proposed in this paper.This ensemble model uses ARIMA model to capture the linear feature of the time series and LSSVM model to fit the nonlinear component of the time series to obtain the synergetic forecasting results by using LSSVM.The validity of the proposed model has been examined by forecasting GDP of our country.Compared with the traditional forecasting methods and the other popular ensemble forecasting method,the result of the presented method is more accurate.