An adjusted Grubbs' and generalized extreme studentized deviation

Abstract

Detecting outlier data is an interesting subject in the statistical field. Grubbs’ test is one of the common detection methods of outlier observation at univariate data sets. This approach is based on the mean and standard deviation of univariate data, and hence, these data are highly affected by the presence of outliers. An improvement for Grubbs’ test is proposed in this paper to increase the power of detecting data.

Keywords

• robust estimation
• detmcd
• dets
• detmm linear discriminant analysis
• 62h30
• 62p03
• 62p05