We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.ST

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Mathematics > Statistics Theory

Title: A Lynden-Bell integral estimator for extremes of randomly truncated data

Authors: Julien Worms (LM-Versailles), Rym Worms (LAMA)
Abstract: This work deals with the estimation of the extreme value index and extreme quantiles for heavy tailed data,randomly right truncated by another heavy tailed variable. Under mild assumptions and the condition thatthe truncated variable is less heavy-tailed than the truncating variable, asymptotic normality is proved for bothestimators. The proposed estimator of the extreme value index is an adaptation of the Hill estimator, in thenatural form of a Lynden-Bell integral. Simulations illustrate the quality of the estimators under a variety ofsituations.
Subjects: Statistics Theory (math.ST)
Cite as: arXiv:1507.04189 [math.ST]
  (or arXiv:1507.04189v1 [math.ST] for this version)

Submission history

From: Julien Worms [view email]
[v1] Wed, 15 Jul 2015 12:35:15 GMT (33kb,D)

Link back to: arXiv, form interface, contact.