On the bootstrap methodology for the estimation of the tail sample fraction

Frederico Almeida Gião Gonçalves Caeiro

On the bootstrap methodology for the estimation of the tail sample fraction

DM - Departamento de Matemática

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

In statistics of extremes we are usually interested in the estimation of pa- rameters of extreme events. Such estimation is usually based on the largest k + 1 order statistics or on the excesses over a high level u. In this paper, we consider the adaptive estimation of either k or u through the nonparametric bootstrap methodology. We shall introduce an improved version of Hall’s bootstrap methodology and compare it with the double bootstrap methodology. The comparison of such methodologies is performed for simulated data sets.

Original language	Unknown
Title of host publication	Proceedings of COMPSTAT 2014
Pages	545-552
Publication status	Published - 1 Jan 2014
Event	COMPSTAT 2014: 21th International Conference on Computational Statistics - Duration: 1 Jan 2014 → …

Conference

Conference	COMPSTAT 2014: 21th International Conference on Computational Statistics
Period	1/01/14 → …

Cite this

@inproceedings{de2932a98b7146c99e90b56e81abd52a,

title = "On the bootstrap methodology for the estimation of the tail sample fraction",

abstract = "In statistics of extremes we are usually interested in the estimation of pa- rameters of extreme events. Such estimation is usually based on the largest k + 1 order statistics or on the excesses over a high level u. In this paper, we consider the adaptive estimation of either k or u through the nonparametric bootstrap methodology. We shall introduce an improved version of Hall{\textquoteright}s bootstrap methodology and compare it with the double bootstrap methodology. The comparison of such methodologies is performed for simulated data sets.",

keywords = "tail sample fraction, extreme value index, heavy tail, bootstrap methodology",

author = "Caeiro, {Frederico Almeida Gi{\~a}o Gon{\c c}alves}",

year = "2014",

month = jan,

day = "1",

language = "Unknown",

isbn = "978-2-8399-1347-8",

pages = "545--552",

booktitle = "Proceedings of COMPSTAT 2014",

note = "null ; Conference date: 01-01-2014",

}

TY - GEN

T1 - On the bootstrap methodology for the estimation of the tail sample fraction

AU - Caeiro, Frederico Almeida Gião Gonçalves

PY - 2014/1/1

Y1 - 2014/1/1

N2 - In statistics of extremes we are usually interested in the estimation of pa- rameters of extreme events. Such estimation is usually based on the largest k + 1 order statistics or on the excesses over a high level u. In this paper, we consider the adaptive estimation of either k or u through the nonparametric bootstrap methodology. We shall introduce an improved version of Hall’s bootstrap methodology and compare it with the double bootstrap methodology. The comparison of such methodologies is performed for simulated data sets.

AB - In statistics of extremes we are usually interested in the estimation of pa- rameters of extreme events. Such estimation is usually based on the largest k + 1 order statistics or on the excesses over a high level u. In this paper, we consider the adaptive estimation of either k or u through the nonparametric bootstrap methodology. We shall introduce an improved version of Hall’s bootstrap methodology and compare it with the double bootstrap methodology. The comparison of such methodologies is performed for simulated data sets.

KW - tail sample fraction

KW - extreme value index

KW - heavy tail

KW - bootstrap methodology

M3 - Conference contribution

SN - 978-2-8399-1347-8

SP - 545

EP - 552

BT - Proceedings of COMPSTAT 2014

Y2 - 1 January 2014

ER -