摘要
目的探讨可以同时抵抗回归数据中出现的Y与X空间的多个异常点的稳健回归估计方法。方法介绍了M估计与LTS估计,引入失效点的概念以讨论其稳健性尺度,并对两个模拟数据进行了分析比较。结果LTS估计优于M估计,失效点较高。结论LTS估计可同时抵抗X与Y方面的异常。
Objective We will explore some robust regression estimates to resist multiple outliers from both sides.Methods We introduce M estimator and LTS estimator,with breakdown point to discuss their robustness.Also,we compare two simulated data as demonstration.Results LTS estimator is superior to M estimator,and has higher breakdown point.Conclusion LTS regression can resist outliers from both sides,and it is more robust.
出处
《中国卫生统计》
CSCD
北大核心
1999年第2期79-81,共3页
Chinese Journal of Health Statistics
基金
全国统计科学研究项目
山西省归国人员基金