大坝变形监测数据回归分析中的因子选择
The Independent Variablas-snooping in the Regression Analysis of Dam Deformation
-
摘要: 本文对我国大坝变形资料分析中常用的逐步回归分析进行了探讨,指出这种方法通常所选的因子数偏少,其原因来自自变量之间的复共线性的影响。为了克服复共线性对因子筛选的影响,文中对因子筛选提出了若干有益的建议。Abstract: In this paper, the regression analysis, which is usually used in dam deformation analysis is discussed. It is shown that the number of selected independent rariables is less than it is real one,because of multi-collinearity between independent variables. In order to over-come effect by multi-collinearity, the author proposes some suggestions about independent variables-snooping.