A Goodness-of-fit Test for the Distribution Tail
Abstract
In order to check that a parametric model provides acceptable tail approximations, we present a test which compares the parametric estimate of an extreme upper quantile with its semiparametric estimate obtained by extreme value theory. To build this test, the sampling variations of these estimates are approximated through parametric bootstrap. Numerical Monte Carlo simulations explore the covering probability and power of the test. A real-data study illustrates these results.
Domains
Methodology [stat.ME]
Origin : Files produced by the author(s)
Loading...