摘要
为了解决OLS估计崩溃点低的问题,系统地介绍了最小截平方和(Least Trimmed Squares,LTS)估计方法及其各种统计特征,并应用LTS回归技术对我国区域经济增长进行了实证分析,结果表明LTS回归技术确实增强了回归的显著性,因此LTS回归技术的应用非常值得推荐.
Aimed at the low value of breakdown point for OLS, the technique of least trimmed squares (LTS) estimation and its various statistic properties are introduced. Then the regional economic growth in our country is analyzed by applying the LTS regression technique. The conclusion indicates that the LTS regression technique assuredly improves the significance of regression. Thus its application is strongly recommended.
出处
《哈尔滨工业大学学报》
EI
CAS
CSCD
北大核心
2009年第4期267-269,共3页
Journal of Harbin Institute of Technology
基金
安徽省自然科学基金资助项目(090416241)
关键词
高崩溃点估计
稳健回归
离群点
杠杆点
最小截平方和
最小协方差矩阵
high breakdown point estimation
robust regression
outliers
leverage point
least trimmed squares (LTS)
minimum covariance matrix (MCD)