BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Generalized Pareto regression trees for extreme event analysis
DTSTART;VALUE=DATE-TIME:20220517T070000Z
DTEND;VALUE=DATE-TIME:20220517T080000Z
DTSTAMP;VALUE=DATE-TIME:20220520T174628Z
UID:indico-event-7900@indico.math.cnrs.fr
DESCRIPTION:Generalized Pareto regression trees for extreme event analysis
\n\nWe provide finite sample results to assess the consistency of Generali
zed Pareto regression trees\, as tools to perform extreme value regression
. The results that we provide are obtained from concentration inequalities
\, and are valid for a finite sample size\, taking into account a misspec
ification bias that arises from the use of a ``Peaks over Threshold'' appr
oach. The properties that we derive also legitimate the pruning strategies
(i.e. the model selection rules) used to select a proper tree that achie
ves compromise between bias and variance. The methodology is illustrated t
hrough a simulation study\, and a real data application in insurance again
st natural disasters.\n\njoint work with S. Farkas\, A. Heranval and O. Lo
pez.\n\n \n\n \n\nA distance : https://univ-lyon1.webex.com/univ-lyon1/
j.php?MTID=m5d77c11315452ecb58aa3988d4b08a98\n\nhttps://indico.math.cnrs.f
r/event/7900/
LOCATION:La doua\, bâtiment Braconnier Salle Fokko du Cloux au 1er étage
URL:https://indico.math.cnrs.fr/event/7900/
END:VEVENT
END:VCALENDAR