Analysis of occurrence of extremes in a time series with a trend

0364201 - ÚTIA 2012 RIV CZ eng C - Konferenční příspěvek (zahraniční konf.)
Volf, Petr
Analysis of occurrence of extremes in a time series with a trend.
Proccedengs of the 29th International Conference on Mathematical Methods in Economics 2011. Praha: Professional Publishing Praha, 2011 - (Dlouhý, M.; Skočdopolová, V.), s. 751-756. ISBN 978-80-7431-059-1.
[29th International Conference on Mathematical Methods in Economics 2011. Jánská Dolina (SK), 06.09.2011-09.09.2011]
Grant CEP: GA ČR GAP402/10/0956
Výzkumný záměr: CEZ:AV0Z10750506
Klíčová slova: extremal value * regression * prediction
Kód oboru RIV: BB - Aplikovaná statistika, operační výzkum
http://library.utia.cas.cz/separaty/2011/SI/volf-analysis of occurrence of extremes in a time series with a trend.pdf

We consider a random series of values and are interested in the analysis and modeling the occurrence of extremes. One of approaches is the analysis of sequence of block maxima. As we assume that the series has a trend, we first select a proper regression model for the block maxima development. From it, a Markov chain of the sequence of extremes is derived. As the transition probabilities of the chain are not tractable analytically, we use the Monte Carlo generation of the chain behavior. Then, from the sample representing the series of block maxima we obtain the rediction of their future behavior.
Trvalý link: http://hdl.handle.net/11104/0199742