Citation
Midi, Habshah and Bagheri, Arezoo
(2013)
Robust multicollinearity diagnostic measures based on minimum covariance determinants approach.
Economic Computation and Economic Cybernetics Studies and Research, 47 (4).
pp. 71-86.
ISSN 0424-267X; ESSN: 1842-3264
Abstract
The classical multicollinearity diagnostic measures are not resistant to high leverage points since their formulation are based on eigen analysis of classical correlation matrix that is very sensitive to the presence of these leverages. The existing robust multicollinearity diagnostics also are not able to diagnose the variables which are collinear to each other. In this paper, we proposed robust multicollinearity diagnostic measures based on the Minimum Covariance Determination (MCD), which is a highly robust estimator of multivariate location and scatter. The results of numerical example and simulation study confirmed the merit of our new proposed robust multicollinearity diagnostic measures.
Download File
Official URL or Download Paper: http://www.ecocyb.ase.ro/Articles20134.htm
|
Additional Metadata
| Item Type: | Article |
|---|---|
| Divisions: | Faculty of Science Institute for Mathematical Research |
| Publisher: | Academy of Economic Studies |
| Keywords: | Multicollinearity; Condition number; Variance Inflation Factor; Variance Decomposition Properties; High leverage points; Minimum covariance determination method |
| Depositing User: | Nurul Ainie Mokhtar |
| Date Deposited: | 31 Dec 2015 04:57 |
| Last Modified: | 19 Sep 2018 01:34 |
| URI: | http://psasir.upm.edu.my/id/eprint/35288 |
| Statistic Details: | View Download Statistic |
Actions (login required)
 |
View Item |