Salvato in:
Dettagli Bibliografici
Autori principali: Kivaranovic, Danijel, Leeb, Hannes
Natura: Preprint
Pubblicazione: 2020
Soggetti:
Accesso online:https://arxiv.org/abs/2007.12448
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910337773600768
author Kivaranovic, Danijel
Leeb, Hannes
author_facet Kivaranovic, Danijel
Leeb, Hannes
contents We show that two popular selective inference procedures, namely data carving (Fithian et al., 2017) and selection with a randomized response (Tian et al., 2018b), when combined with the polyhedral method (Lee et al., 2016), result in confidence intervals whose length is bounded. This contrasts results for confidence intervals based on the polyhedral method alone, whose expected length is typically infinite (Kivaranovic and Leeb, 2020). Moreover, we show that these two procedures always dominate corresponding sample-splitting methods in terms of interval length.
format Preprint
id arxiv_https___arxiv_org_abs_2007_12448
institution arXiv
publishDate 2020
record_format arxiv
spellingShingle A (tight) upper bound for the length of confidence intervals with conditional coverage
Kivaranovic, Danijel
Leeb, Hannes
Methodology
We show that two popular selective inference procedures, namely data carving (Fithian et al., 2017) and selection with a randomized response (Tian et al., 2018b), when combined with the polyhedral method (Lee et al., 2016), result in confidence intervals whose length is bounded. This contrasts results for confidence intervals based on the polyhedral method alone, whose expected length is typically infinite (Kivaranovic and Leeb, 2020). Moreover, we show that these two procedures always dominate corresponding sample-splitting methods in terms of interval length.
title A (tight) upper bound for the length of confidence intervals with conditional coverage
topic Methodology
url https://arxiv.org/abs/2007.12448