Publication detail

Extreme value estimation for correlated observations

HOLEŠOVSKÝ, J. FUSEK, M. MICHÁLEK, J.

Czech title

Odhady extrémních hodnot pro korelovaná pozorování

English title

Extreme value estimation for correlated observations

Type

conference paper

Language

en

Original abstract

Statistical modeling of extreme events is the object of interest in many application areas. When estimating such rare events from a time series, extreme value theory is commonly used. In that case, series with independent members are required. However, the assumption of independence is not satisfied in many situations. There are two approaches (block maxima, peaks-over-threshold) which result in series with independent members, but the length of the series is substantially reduced. In this paper, stationary series with short-time dependence described by the extremal index theta is considered, and two estimators of theta are introduced. Behavior of the estimators is assessed using simulations. The described methods are used in an analysis of real hydrological data, and compared with classical peaks-over-threshold approach.

Czech abstract

Statistické modelování extrémních hodnot patří k významným oblastem mnoha vědních odvětví. Odhady frekvencí výskytů těchto řídkých událostí pro časové řady jsou běžně postaveny na teorii extrémních hodnot, která ovšem využívá nezávislých pozorování. Tento požadavem ovšem často není splněn. V praktických situacích jsou používány dva přístupy (bloková maxima a peaks-over-threshold), pomocí nichž je původní řada zredukována (často významně) na řadu s vzájemně nezávislými pozorováními. V příspěvku uvažujeme stacionární řady s "krátkodobou závislostí", která může být popsána pomocí parametru tzv. extremálního indexu. Jsou představeny dva odhady extremálního indexu a jejich chování je studováno za pomocí simulací. Získané výsledky jsou aplikovány k analýze reálných hydrologických dat a také srovnány s přístupem peaks-over-threshold.

English abstract

Statistical modeling of extreme events is the object of interest in many application areas. When estimating such rare events from a time series, extreme value theory is commonly used. In that case, series with independent members are required. However, the assumption of independence is not satisfied in many situations. There are two approaches (block maxima, peaks-over-threshold) which result in series with independent members, but the length of the series is substantially reduced. In this paper, stationary series with short-time dependence described by the extremal index theta is considered, and two estimators of theta are introduced. Behavior of the estimators is assessed using simulations. The described methods are used in an analysis of real hydrological data, and compared with classical peaks-over-threshold approach.

Keywords in Czech

rozdělení extrémních hodnot, extremální index, peaks over threshold, stacionární proces

Keywords in English

extreme value distribution, extremal index, peaks over threshold, stationary process

RIV year

2014

Released

25.06.2014

Publisher

Brno University of Technology, Faculty of Mechanical Engineering, Institute of Automation and Computer Science

Location

Brno, Czech Republic

ISBN

978-80-214-4984-8

ISSN

1803-3814

Book

Mendel 2014 20th International Conference of Soft Computing

Pages from–to

359–364

Pages count

6

BIBTEX


@inproceedings{BUT108396,
  author="Jan {Holešovský} and Michal {Fusek} and Jaroslav {Michálek},
  title="Extreme value estimation for correlated observations",
  booktitle="Mendel 2014 20th International Conference of Soft Computing",
  year="2014",
  month="June",
  pages="359--364",
  publisher="Brno University of Technology, Faculty of Mechanical Engineering, Institute of Automation and Computer Science",
  address="Brno, Czech Republic",
  isbn="978-80-214-4984-8",
  issn="1803-3814"
}