International Journal of Operations Research and Optimization
8:10:45
 
Welcome to AACS
 
More....

International Journal of Operations Research and OptimizationJul-Dec 2016 Vol:7 Issue:2

Detecting influential observations using Least Absolute Deviations Regression

Abstract

Robust regression is a form of regression analysis designed to circumvent some limitation of traditional parametric and Non-parametric methods. Robust methods are known as resistant of abnormal values and other violations of model assumptions and appropriate for a broad category of distributions. It is an alternative to lease squares regression when data are contaminated with outliers or influential observations. It can be also used for the purpose of detecting influential observations. Detecting influential observations using Least Absolute Deviations Regression are designed to be not overly affected by violation of assumption by the underlying data generating process. In this paper it is proposed to compare LAD method with the iteratively reweighted least square and ordinarily least square method by detecting influential observations. It is observed that LAD method is more efficient in estimating the parameters in all cases, the distribution of errors follows heavy tailed distributions and in the case of contamination of the data with abnormal values. The LAD regression methods are not only robust but give consistent results in detecting outliers.

Author

B. Ashok and R. Elangovan  ( Pages 83-92 )
Email:
Affiliation:      DOI:

Keyword

Deviations Regression, LAD method, Influential observations

References

Download paper
AACS Journals
Visitor:-
Copyright © 2020 AACS All rights reserved