期刊名称：Çukurova Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi
印刷版ISSN：1300-3747
出版年度：2005
卷号：9
期号：1
页码：1-21
语种：English
出版社：Çukurova University
摘要：Numerous methods have been suggested for outier detection in multivariatedata. In this paper we propose a new method for robust principle component analysis.The main idea is to obtain robust principal components based on robust covariance estimate of the clean data obtined from BACON algorithm (Billor et al., 2000.0 Thismethlod yields sgnifcant componcnts that are free of outiers and that contin most of
the information in a data matrix.The aplpcabiliti of this method is shown by using two diferent data sets.
其他摘要：Qok degiskenli veri kimelerinde sapan degerlerin belirlenmesi igin pek cokyontem onerilmistir. Bu konuda son yillarda onerilen yontemlerden BACONalgoritmas1 buyuk veri kimeleri icin hesapsal olarak etkin ve maskeleme ve swamping problemlerine kars1 dayank
