Data Analysis Papers
Robust Regression and Outlier Detection with the ROBUSTREG
Procedure
Colin Chen, SUGI Proceedings, 2002.
Abstract
Robust regression is an important tool for analyzing data that are contaminated with outliers. It can be used to detect outliers and to provide resistant (stable) results in the presence of outliers. This paper introduces the ROBUSTREG procedure, which is experimental in SAS/STATâ Version 9. The ROBUSTREG procedure implements the most commonly used robust regression techniques. These include M estimation (Huber, 1973), LTS estimation (Rousseeuw, 1984), S estimation (Rousseeuw and Yohai, 1984), and MM estimation (Yohai, 1987). The paper will provide an overview of robust regression methods, describe the syntax of PROC ROBUSTREG, and illustrate the use of the procedure to fit regression models and display outliers and leverage points. This paper will also discuss scalability of the ROBUSTREG procedure for applications in data cleansing and data mining.