Developing a New Estimator in Linear Regression Model

Adewale, F. Lukman and Kayode, Ayinde and Alabi, Olatayo and Rasaq, Bamidele and Benedicta, Aladeitan and Theophilus, Adagunodo (2019) Developing a New Estimator in Linear Regression Model. Journal of Physics: Conference Series, 1299 (012128). pp. 1-12.

[img] Text
Lukman_2019_J._Phys.__Conf._Ser._1299_012128-1.pdf - Published Version

Download (1MB)
Official URL:


Abstract. The most popularly used method of estimating the parameters in a linear regression model is the Ordinary Least Squares (OLS) Estimator. This estimator is considered best when certain assumptions are satisfied. However, when any of these assumptions fail there is a need for other estimators. In this article, we proposed a new estimator that can handle jointly the violations of three of these assumptions which include multicollinearity, outliers, and an auto-correlated error term. Three estimators were combined to form Generalized-Ridge-Lad estimator. We compared the performance of the new estimator with some of the existing estimators in terms of their mean square error. The proposed estimator (GLSRIDGELAD) perform consistently better than other estimators when the three problems exist.

Item Type: Article
Uncontrolled Keywords: Multicollinearity, Autocorrelated error, Outliers, Generalized ridge, LAD-estimator
Subjects: Q Science > QC Physics
Divisions: Faculty of Law, Arts and Social Sciences > School of Humanities
Depositing User: Dr. Bamidele Rasak
Date Deposited: 18 Nov 2019 09:16
Last Modified: 18 Nov 2019 09:16

Actions (login required)

View Item View Item