An alternative estimator for regression coefficients with outliers
Issued Date
2010
Issued Date (B.E.)
2553
Available Date
Copyright Date
Resource Type
Series
Edition
Language
eng
File Type
application/pdf
No. of Pages/File Size
x, 128 leaves : ill. ; 30 cm.
ISBN
ISSN
eISSN
Other identifier(s)
Identifier(s)
Access Rights
Access Status
Rights
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Rights Holder(s)
Physical Location
National Institute of Development Administration. Library and Information Center
Bibliographic Citation
Citation
Titirut Mekbunditkul (2010). An alternative estimator for regression coefficients with outliers. Retrieved from: http://repository.nida.ac.th/handle/662723737/383.
Title
An alternative estimator for regression coefficients with outliers
Alternative Title(s)
Author(s)
Advisor(s)
Editor(s)
item.page.dc.contrubutor.advisor
Advisor's email
Contributor(s)
Contributor(s)
Abstract
An important problem often found in a regression analysis is that a structural change in regression exists and/or the observed data contain outliers. These can lead to a violation of the Gauss-Markov assumptions and affect least squares (LS), rendering the regression inadequate. In this dissertation, an alternative regression model has been constructed from a combination of two principle ideas, namely the Tobit and piecewise regressions. This combined model, called the Tobit-piecewise (TP) regression model, can be suitably applied to cope with the above problem in the sense that it can reduce the effect of outliers by down-weighting them. A suitable likelihood function is derived under the desired conditions so that the TP estimator is verified as the MLE of the parameter vector in the TP regression model. The estimator is expressed in the following formula: Properties, such as the bias and the asymptotic variance-covariance matrix of the TP estimator are verified. At the beginning of the study, the parameter is assumed to be known, but in many applications where this is not desirable, an estimator based on an ML estimation is provided. Simulation studies are applied to support the theoretical results and the following conclusions are obtained. 1) The TP estimator is biased and it can reduce the effect of a structural change in the regression analysis and/or the existence of outliers. In addition, the TP regression model can preferably represent the relationships between the data when compared with other models, namely LS, Tobit and piecewise. 2) When data consist of two structures and/or outliers in the x-direction, TP is identical to piecewise and can be appropriately utilized. 3) When data consist of y- or xy-direction outliers but do not consist of two or more structures, TP is the same as Tobit and they are considered adequate. 4) When data do not contain outliers and do not consist of two or more structures, TP is no different from LS and they are both BLUE.
Table of contents
Description
Thesis (Ph.D. (Statistics))--National Institute of Development Administration, 2010