Robust Estimation for Parameters of the Extended Burr Type III Distribution

MERT KANTAR Y., Yildirim V.

COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION, vol.44, no.7, pp.1901-1930, 2015 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 44 Issue: 7
  • Publication Date: 2015
  • Doi Number: 10.1080/03610918.2013.839032
  • Journal Name: COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.1901-1930
  • Keywords: Outliers, Quantile estimator, Robust regression estimator, The extended Burr Type III distribution, WEIBULL DISTRIBUTION, REGRESSION, LOCATION, PARETO, MODEL
  • Anadolu University Affiliated: Yes

Abstract

We consider various robust estimators for the extended Burr Type III (EBIII) distribution for complete data with outliers. The considered robust estimators are M-estimators, least absolute deviations, Theil, Siegel's repeated median, least trimmed squares, and least median of squares. Before we perform the aforementioned estimators for the EBIII, we adapt the quantiles method to the estimation of the shape parameter k of the EBIII. The simulation results show that the considered robust estimators generally outperform the existing estimation approaches for data with upper outliers, with certain of them retaining a relatively high degree of efficiency for small sample sizes.