Title: Robust Location and Scatter Estimators in Multivariate Analysis
Abstract: Frontiers in Statistics, pp. 467-490 (2006) No AccessRobust Location and Scatter Estimators in Multivariate AnalysisYijun ZuoYijun ZuoDepartment of Statistics and Probability, Michigan State University, East Lansing, MI 48824, USAhttps://doi.org/10.1142/9781860948886_0021Cited by:3 PreviousNext AboutSectionsPDF/EPUB ToolsAdd to favoritesDownload CitationsTrack CitationsRecommend to Library ShareShare onFacebookTwitterLinked InRedditEmail Abstract: The sample mean vector and the sample covariance matrix are the corner stone of the classical multivariate analysis. They are optimal when the underlying data are normal. They, however, are notorious for being extremely sensitive to outliers and heavy tailed noise data. This article surveys robust alternatives of these classical location and scatter estimators and discusses their applications to the multivariate data analysis. FiguresReferencesRelatedDetailsCited By 3Robust multivariate functional discriminant coordinatesMirosław Krzyśko and Łukasz Smaga15 July 2019 | Communications in Statistics - Simulation and Computation, Vol. 49, No. 3Robustness of Msplit(q) estimation: A theoretical approachRobert Duchnowski and Zbigniew Wiśniewski2 July 2019 | Studia Geophysica et Geodaetica, Vol. 63, No. 3Adaptive Exponential Power Depth with Application to ClassificationYunlu Jiang, Canhong Wen and Xueqin Wang1 October 2018 | Journal of Classification, Vol. 35, No. 3 Frontiers in StatisticsMetrics Downloaded 8 times History PDF download