多元门限回归模型的一种建模方法

Method of Building a Multivariate Threshold Regression Model

本文根据门限自回归模型的基本思想[1]，提出一种多元门限回归模型的建模方法。其特点是充分考虑了预报系统中某些特殊预报因子突变点对预报关系的改变作用。数值实例表明，该模型在模拟和预报精度上比一般线性逐步回归模型有一定程度的提高。

In this paper, a method of building multivariate threshold regression model is given. The idea is mainly based on the fact that the relation between predictor and predictand will change when some spe-cial predictors change abruptly. Computational results have shown better effectiveness using multivariate threshold regression model than using general regression models.