2015•arXiv (Cornell University)Open access

A comparative review of generalizations of the extreme value distribution

Eliane Cantinho Pinheiro, Silvia L. P. Ferrari

Open full text 2 citations

Abstract

The extreme value distribution, also known as the Gumbel distribution, is widely applied for extreme value analysis but has certain drawbacks in practice because it is a non heavy-tailed distribution and is characterized by constant skewness and kurtosis. Our goal is to present a literature review of the distributions that contain the extreme value distribution embedded in them and to identify those that have flexible skewness and kurtosis and those that are heavy-tailed. The generalizations of the extreme value distribution are described and compared using an application to a wind speed data set and Monte Carlo simulations. We show that some distributions suffer from overparameterization and coincide with other generalized Gumbel distributions with a smaller number of parameters, i.e., are non-identifiable. Our study suggests that the generalized extreme value distribution and a mixture of two extreme value distributions should be considered in practical applications.

About this research paper

What this paper is about

The extreme value distribution, also known as the Gumbel distribution, is widely applied for extreme value analysis but has certain drawbacks in practice because it is a non heavy-tailed distribution and is characterized by constant skewness and kurtosis. Our goal is to present a literature review of the distributions that contain the extreme value distribution embedded in them and to identify those that have flexible skewness and kurtosis and those that are heavy-tailed. The generalizations of the extreme value distribution are described and compared using an application to a wind speed data set and Monte Carlo simulations. We show that some distributions suffer from overparameterization and coincide with other generalized Gumbel distributions with a smaller number of parameters, i.e., are non-identifiable. Our study suggests that the generalized extreme value distribution and a mixture of two extreme value distributions should be considered in practical applications.

Why it matters

OpenAlex reports 2 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

The extreme value distribution, also known as the Gumbel distribution, is widely applied for extreme value analysis but has certain drawbacks in practice because it is a non heavy-tailed distribution and is characterized by constant skewness and kurtosis. Our goal is to present a literature review of the distributions that contain the extreme value distribution embedded in them and to identify those that have flexible skewness and kurtosis and those that are heavy-tailed. The generalizations of the extreme value distribution are described and compared using an application to a wind speed data set and Monte Carlo simulations. We show that some distributions suffer from overparameterization and coincide with other generalized Gumbel distributions with a smaller number of parameters, i.e., are non-identifiable. Our study suggests that the generalized extreme value distribution and a mixture of two extreme value distributions should be considered in practical applications.

Key concepts: Gumbel distribution, Kurtosis, Extreme value theory, Generalized extreme value distribution, Skewness, Mathematics, Distribution (mathematics), Statistical physics

Related papers

Back to paper searchBrowse research topicsOriginal source
A comparative review of generalizations of the extreme value distribution — Research Paper | ScholarLens