Combining the Liu-type estimator and the principal component regression estimator

Inan D.

STATISTICAL PAPERS, vol.56, no.1, pp.147-156, 2015 (SCI-Expanded) identifier

  • Publication Type: Article / Article
  • Volume: 56 Issue: 1
  • Publication Date: 2015
  • Doi Number: 10.1007/s00362-013-0571-5
  • Journal Name: STATISTICAL PAPERS
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.147-156
  • Keywords: Liu-type estimator, r-k class estimators, Principal component regression estimator, Ridge estimator, Multicollinearity, RIDGE-REGRESSION
  • Marmara University Affiliated: Yes

Abstract

In this study a new two-parameter estimator which includes the ordinary least squares, the principal components regression (PCR) and the Liu-type estimator is proposed. Conditions for the superiority of this new estimator over the PCR, r-k class estimator and Liu-type estimator are derived. Furthermore the performance of this estimator is compared with the other estimators in different conditions with simulation studies.