George Mason University
CSI/Statistics Colloquium Series
Seminar Announcement


The Minimum Sum of Absolute Errors Regression: Some Recent Results

Subhash C. Narula

Virginia Commonwealth University


ABSTRACT

It has been shown that the minimum sum of absolute errors regression is more robust than the least squares regression for some types of outliers. Further it has been shown that even if the value of certain variable for an observation is changed with in limits, it leaves the fitted minimum sum of absolute errors regression unchanged. I will discuss this special property of the minimum sum of absolute errors regression for the simple and the multiple linear regression and illustrate the results with examples.


Friday, November 6, 1998
George W. Johnson Center, Assembly Room G
Seminar at 10:45 a.m.
Refreshments at 10:30 a.m.