An Evaluation of Some Robust Estimators of Regression Coefficients

Date
2015-12
Authors
Ding, Jie
Journal Title
Journal ISSN
Volume Title
Publisher
Faculty of Graduate Studies and Research, University of Regina
Abstract

In the theory of regression analysis, the method of least squares is most commonly used because of its mathematical beauty and computational simplicity. However, this method is now criticized more and more because it often has very poor performance when there are outliers in the data. In this connection a variety of robust statistics are developed for that they are not unduly a ected by outliers. In this thesis comparison studies have been made for several robust statistics to see which performs better than the others. Monte Carlo simulation has been used to carry out the comparison of these statistics, including the least absolute deviations estimator and the least median of squares estimator, least trimmed squares estimator and some M-estimators.

Description
A Thesis Submitted to the Faculty of Graduate Studies and Research In Partial Fulfillment of the Requirements for the Degree of Master of Science in Statistics, University of Regina. vi, 59 p.
Keywords
Citation
Collections