Comparison of different estimation methods for extreme value distribution

Yılmaz A., Kara M., Özdemir O.

JOURNAL OF APPLIED STATISTICS, vol.48, no.13-15, pp.2259-2284, 2021 (SCI-Expanded) identifier identifier identifier

  • Publication Type: Article / Article
  • Volume: 48 Issue: 13-15
  • Publication Date: 2021
  • Doi Number: 10.1080/02664763.2021.1940109
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, ABI/INFORM, Aerospace Database, Business Source Elite, Business Source Premier, CAB Abstracts, Veterinary Science Database, zbMATH
  • Page Numbers: pp.2259-2284
  • Keywords: Bayesian methods, Extreme value distribution, L-moments estimators, Monte Carlo simulation, new estimators, Trimmed L-moments estimators
  • Van Yüzüncü Yıl University Affiliated: Yes


The extreme value distribution was developed for modeling extreme-order statistics or extreme events. In this study, we discuss the distribution of the largest extreme. The main objective of this paper is to determine the best estimators of the unknown parameters of the extreme value distribution. Thus, both classical and Bayesian methods are used. The classical estimation methods under consideration are maximum likelihood estimators, moment's estimators, least squares estimators, and weighted least squares estimators, percentile estimators, the ordinary least squares estimators, best linear unbiased estimators, L-moments estimators, trimmed L-moments estimators, and Bain and Engelhardt estimators. We also propose new estimators for the unknown parameters. Bayesian estimators of the parameters are derived by using Lindley's approximation and Markov Chain Monte Carlo methods. The asymptotic confidence intervals are considered by using maximum likelihood estimators. The Bayesian credible intervals are also obtained by using Gibbs sampling. The performances of these estimation methods are compared with respect to their biases and mean square errors through a simulation study. The maximum daily flood discharge (annual) data sets of the Meric River and Feather River are analyzed at the end of the study for a better understanding of the methods presented in this paper.