变量具误差的多元线性回归不确定度评定方法

Method of evaluating uncertainties of multivariate linear regression for variables with errors

给出了多元线性回归中常数项参数标准差计算的一种新方法,在把自变量误差影响转化为因变量不确定度的一个B类分量的情况下,导出了自变量含误差时多元线性回归参数标准不确定度的评定方法,并给出了一元和二元情况的简便计算形式,还给出了对因变量或自变量非随机误差限影响作取舍处理的一种判定方法.仿真计算取得好的效果.

A new method of computing the standard deviation of the constant term parameter of the multivariate linear regression is presented.On the condition that the effects of the error of the independent variables are transformed into a Type B component of the uncertainty of the dependent variable,the method of evaluating the standard uncertainty of the multivariate linear regression parameter for the independent variables with errors is presented,and the simple forms of one-variable and two-variable are obtained....