作者: Marie Ng , Rand R. Wilcox
DOI: 10.22237/JMASM/1272686640
关键词:
摘要: Numerous multivariate robust measures of location have been proposed and many found to be unsatisfactory in terms their small-sample efficiency. Several new recently derived, however, nothing is known about efficiency or how they compare the sample mean under normality. This research compared for p = 2, 5, 8 with sizes n 20 50 p-variate data. Although previous studies indicate that so-called skipped estimators are efficient, this study variations approach can perform poorly when small exceeds 5. One best was a estimator where outliers detected by projection method eliminated. The TBS, OGK RMBA were included and; some cases, performed well, serious exceptions identified suggesting based on projection-type outlier detection preferable