摘要
提出了回归分析中异性方差的检验和对估计结果带来的影响以及消除其异方差性的两种方法：对原模型进行变换和加权最小二乘法．
The article presents two methods in retum analysis resulting from changing original modal and weighted least squares: (i) singular Varuance brings about effects on estimative results; (ii)How to eleminaten its difference.
关键词
回归模型
最小二乘法
随机误差
异性方差
Return model
Least squares
Singular variance
Error at random
