一元加权最小二乘估计处理异方差性及SAS实现

Simple Weighted Least Squares on Dealing with Heteroscedasticity and its SAS Programme

分析线性回归分析中异方差性的产生原因和影响,介绍加权最小二乘估计法处理一元回归分析中异方差性的原理,并用SAS软件加以实现。通过比较SAS软件输出结果,得出当存在异方差性时,加权最小二乘估计法比普通最小二乘估计法估计效果更好的结论。

Analyze the reason and the influence of heteroscedasticity in linear regression, introduce the theory of Weighted Least Squares method（WLS） on dealing with heteroscedasticity in simple linear regression, and resolve it by programming a SAS programme. By comparing the results of SAS software, we will seen Weighted Least Squares method（WLS） is better than Ordinary Least Squares method（OLS） on linear regression analysis when heteroscedasticity exists.