欢迎访问林业科学,今天是

林业科学 ›› 1999, Vol. 35 ›› Issue (5): 5-11.

• 论文及研究报告 • 上一篇    下一篇

论加权回归与建模

曾伟生,骆期邦,贺东北   

  1. 国家林业局中南调查规划设计院 长沙 410014
  • 收稿日期:1998-04-06 出版日期:1999-09-25 发布日期:1999-09-25

RESEARCH ON WEIGHTING REGRESSION AND MODELLING

Weisheng Zeng,Qibang Luo,Dongbei He   

  1. South-Central Forest Inventory & Planning Institute Changsha 410014
  • Received:1998-04-06 Online:1999-09-25 Published:1999-09-25

摘要:

以加权回归估计方法为核心,对林业上常用模型的异方差性进行了研究,提出了能彻底消除异方差的最佳权函数。并对模型的评价指标进行了探讨,提出了评价通用性回归模型的3大指标,并分析了加权回归估计与这些评价指标之间的关系。最后对样本资料的收集进行了讨论,提出了收集建模样本应遵循的基本原则。

关键词: 加权回归, 建模, 异方差, 模型评价

Abstract:

Taking weighting regression estimation method as the core, the heteroscedasticity of the general models used in forestry was discussed, and an optimal weight function was presented that could completely eliminate the unequal variance.Secondly, the model assessment was studied, and four important indices for commonly-used regression model assessment were presented, and the relationship between the four indices and weighting regression estimation was analysed.Finally, the collection of modelling sample was discussed, and the basic princinple for collecting sample data was presented.

Key words: Weighting regression, Modelling, Heteroscedasticity, M odel assessment