数学物理学报

• 论文 • 上一篇    下一篇

平衡单向分类随机效应模型中的多个异常值检验

石磊; 何利平; 黄梅   

  1. 云南财经大学统计与数学学院, 昆明 650221
  • 收稿日期:2004-12-17 修回日期:2006-01-15 出版日期:2007-06-25 发布日期:2007-06-25
  • 通讯作者: 石磊
  • 基金资助:
    基金项目:国家自然科学基金(10261009)资助

Detection of Mutiple-outliers in Random-effects Model of Balanced One-way Classification

Shi Lei; He Liping; Huang Mei   

  1. Statistics and Mathematics College, Yunnan University of Finance and Economics, Kunming 650221
  • Received:2004-12-17 Revised:2006-01-15 Online:2007-06-25 Published:2007-06-25
  • Contact: Shi Lei

摘要: 该文研究平衡单向分类随机效应模型中多个异常值的检验问题. 在基于随机效应上的均值滑动模型下导出了似然比检验统计量, 并给出了其精确分布及水平异常值的检验过程. 在基于观测误差上的均值滑动模型下,利用得分检验统计量给出了多个异常值的检验过程.

关键词: 平衡单向分类随机效应模型, 多个异常值, 检验, 得分统计量

Abstract: This paper studies the detection of multiple-outliers in balanced random-effects model of one-way classification. The authors derive the likelihood ratio test under the mean-shift model in random-effects. The distribution of test statistic is given and employed to obtain the test procedure. The authors also study the test of mean-shift outlier model in error variables, and a score test is used to identify the multiple-outliers in the data.

Key words: Random-effects model of one-way classification, Mutiple-outliers, Test, Score statistic

中图分类号: 

  • 62J99