摘要
通过添加一个正则化因子α,使时间序列AR(n)模型的最小二乘估计(X′X)-1X′Y变为(X′X+αI)-1X′Y,改善了时间序列分析模型中信息矩阵的病态程度,避免了时间序列分析模型产生不适定;经济统计数据分析表明,新的正则化时间序列分析模型在一定程度上起到了稳定所求参数的作用。
A regularization factor is added to time series AR （n） model that changes least square estimation. The ill-eonditioned matrix is improved in time series analysis model, which avoid the ill-posed of time series analysis model. Analysis of economic example shows that this new regularized time series analysis model played a role in stabilizing the parameters to some extent.
出处
《价值工程》
2010年第2期86-87,共2页
Value Engineering
关键词
正则化
时间序列
AR模型
经济分析
regularization
time series
AR model
economic analysis
