摘要
利用传统曲率理论度量回归模型的非线性程度,分析其缺陷,并对几类实例模型进行探讨,进而提出基于局部近似的新度量方法,在一定程度上弥补了传统方法的缺陷;还研究了模型参数基于局部线性近似的最小二乘估计的渐进性质及其求法中迭代初值的选取和改进方向.
Based on classical methods of the non-linearity property of the non-linearity regression model, analysis of its limitation is made. This paper promotes a new method about local approximation through the discussion for the convergence of some models, which remedy the defect of the traditional ways. The asymptotic property of least square estimator and the method of iterative initial value selection are studied. And a new trend for the method to measure the model＇s non-linearity is given.
出处
《湖南文理学院学报（自然科学版）》
CAS
2013年第1期10-13,共4页
Journal of Hunan University of Arts and Science(Science and Technology)
基金
湖南省自然科学基金资助项目(12JJ6004)
关键词
非线性度量
固有曲率
参数效应曲率
局部曲率
measure of non-linearity
intrinsic curvature
parameter-effects curvature
local curvature
