Computational Procedures for Improving Extreme Value Estimation in Time Series Modelling

Dora Prata Gomes, Clara Cordeiro, Manuela Neves

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In the last decades, some work has been developed in parameter estimation of extreme values jointly with time series analysis. Those results show relevant asymptotic properties. However, for finite samples, limiting results provide approximations that can be poor. Some challenges have been developed by combining Extreme Value Theory and time series modelling to obtain more reliable extreme value parameter estimates. In classical time series modelling a key issue is to determine how many parameters must be included in the model. Special care must be given to extreme events in the series that need specific statistical procedures based on the behaviour of extremes. Resampling procedures such as the jackknife and the bootstrap have been used to improve parameters estimation in Extreme Value Theory combined with time series modelling. New approaches, based on bootstrap procedures are shown and are illustrated with a real data set using the software.
Original languageEnglish
Title of host publicationComputational Science and Its Applications – ICCSA 2023 Workshops
Subtitle of host publicationAthens, Greece, July 3–6, 2023, Proceedings, Part II
EditorsOsvaldo Gervasi, Beniamino Murgante, Ana Maria A. C. Rocha, Chiara Garau, Francesco Scorza, Yeliz Karaca, Carmelo M. Torre
Place of PublicationCham
PublisherSpringer
Pages84-96
Number of pages13
ISBN (Electronic)978-3-031-37108-0
ISBN (Print)978-3-031-37107-3
DOIs
Publication statusPublished - 2023
Event23rd International Conference on Computational Science and Its Applications, ICCSA 2023 - Athens, Greece
Duration: 3 Jul 20236 Jul 2023

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume14105 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference23rd International Conference on Computational Science and Its Applications, ICCSA 2023
Country/TerritoryGreece
CityAthens
Period3/07/236/07/23

Keywords

  • Extreme value theory
  • Parameter estimation
  • Resampling procedures
  • Time series

Fingerprint

Dive into the research topics of 'Computational Procedures for Improving Extreme Value Estimation in Time Series Modelling'. Together they form a unique fingerprint.

Cite this