A Review Of Some Robust Data Analysis And Multiple Outlier Detection Procedures

Abstract

Some recent contributions to robust data analysis and multiple outlier detection are discussed. Two methods of analysis producing robust estimates and sets of weights which may be inspected for outliers are described and compared. Some examples of their application are given to support the recommendation that both ordinary least squares and a robust method of analysis should be part of routine data analysis.

Keywords:

• Linear models
• multiple outliers
• robust analysis