首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于信息量准则的广义Lambda分布变点分析
作者姓名:黄传明  张晓旭  张三国
作者单位:1. 中国科学院大学数学科学学院, 北京 100049; 2. 中国科学院大学管理学院, 北京 100190
基金项目:教育部回国留学人员科研启动基金及中国科学院研究生院院长基金(Y05101AY00)资助
摘    要:研究广义Lambda分布的变点问题,结合信息量准则方法实现了变点检测.前人曾指出,为使变点检测结果更具统计说服力,可引入显著性水平α及相关临界值cα.本文指出,在一般条件下,此时与通常极大似然比检验渐近结果是等价的.给出5种不同信息量准则下模拟数据的检测结果及分析,并将该方法应用于实际数据.实验结果表明,该方法能够有效识别出变点位置.

关 键 词:广义Lambda分布  变点检测  信息量准则  似然比检验  
收稿时间:2013-02-05
修稿时间:2013-04-16

Change point analysis with generalized Lambda distribution based on information criterion
Authors:HUANG Chuan-Ming  ZHANG Xiao-Xu  ZHANG San-Guo
Institution:1. School of Mathematical Sciences, University of Chinese Academy of Sciences, Beijing 100049, China; 2. School of Management, University of Chinese Academy of Sciences, Beijing 100190, China
Abstract:In this article, we study the change point problem with generalized Lambda distribution based on information criterion. It was previously pointed out that, to make the conclusion more statistically convincing, the significant level α and its associated critical value cα could be introduced. We prove that, under ordinary conditions, this kind of information criterion is equivalent to the maximum likelihood ratio test in asymptotic formation. The simulation study is carried out with five different information criterions and the method is applied to treat practical data. The results show the validity of this method.
Keywords:generalized Lambda distribution                                                                                                                        change point analysis                                                                                                                        information criterion                                                                                                                        likelihood ratio test
点击此处可从《》浏览原始摘要信息
点击此处可从《》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号