A Study on Tests of Hypothesis Based on Ridge Estimator

GÖKPINAR E., EBEGİL M.

GAZI UNIVERSITY JOURNAL OF SCIENCE, vol.29, no.4, pp.769-781, 2016 (ESCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 29 Issue: 4
  • Publication Date: 2016
  • Journal Name: GAZI UNIVERSITY JOURNAL OF SCIENCE
  • Journal Indexes: Emerging Sources Citation Index (ESCI), Scopus
  • Page Numbers: pp.769-781
  • Keywords: Biasing parameters, Ridge regression, Hypothesis testing, Type I error rate, Power of test, ORDINARY LEAST-SQUARES, REGRESSION ESTIMATORS, SHRINKAGE ESTIMATORS, SIMULATION, PARAMETERS, MODEL
  • Gazi University Affiliated: Yes

Abstract

The literature of Ridge regression include many articles that deal with point estimation of the coefficients vector. However, few of them tackle the statistical inference problem about or some of its components. One of them is introduced by Halawa and Basuiouni[1] who present non-exact tests based on Ridge regression by using two different biasing parameters (k) which are proposed by Hoerl and Kennard [2] and Hoer et al. [3]. Thus, we investigate others popular k values used the Ridge regression for testing significance of regression coefficients. We compare tests in terms of type I error rates and powers by using Monte Carlo simulation. In addition, a real data example is presented.