多维时序定阶方法对偏最小二乘回归模型预报精度的改进

The Method of Pricing Order of the Multivariate Autoregression Model for Amelioration of Forecast Precision of the Partial Least-Squares Regression Model

运用时间序列多维自回归模型的定阶方法 ,解决了偏最小二乘回归模型中自变量的选择问题 .通过对我国财政收入的预报分析表明 ,这两种统计模型的结合使用 。

This paper makes use of the method o f pricing order of the Multivariate Autoregression Model to resolve the choosing problem of variable about the Partial Least-Squares Regression Model. The indi cation of forecast analysis for China′s fiscal expenditures, the combination of the two kinds of models can raise the forecast precision.