摘要
用B样条函数最小二乘法的非参数回归与时间序列相结合的方法建立了季节性时间序列预测模型.利用滑动平均估计季节项,再利用B样条函数非参数回归估计长期项和周期波动,对于随机项建立ARMA模型,最后对某产品需求量进行了实例分析.结果表明该方法有较高的预测精度.
Nonparametric regression is combined with the method of time series to establish the prediction model of time series with seasonal fluctuations. Seasonal index is first established by the moving average method. The long-time trend＇ and cyclic fluctuation are estimated by nonparametric regression based on smooth spline of the B form. In addition, ARMA mode is established with the random term. The order of a product is finally analysed showing that the method is effective.
出处
《北京理工大学学报》
EI
CAS
CSCD
北大核心
2007年第4期370-373,共4页
Transactions of Beijing Institute of Technology
