Extreme Conditional Quantile Estimation in High Dimensions: A Comparative Study

Laurent Gardès & Alex Podgorny

International Statistical Review2026https://doi.org/10.1111/insr.70023article
AJG 3ABDC A
Weight
0.50

Abstract

Summary This paper addresses the problem of estimating extreme conditional quantiles in high‐dimensional settings. We mainly focus on the case where the conditional distribution is heavy tailed. We consider recent estimation procedures based on machine learning techniques, including neural networks and random forests, as well as dimension reduction approaches specifically designed for extreme values. A comprehensive simulation study evaluates their performance across various scenarios, investigating the influence of the covariate dimension, the second‐order tail behaviour, the complexity of the link between covariate and tail index, the correlation structure of the random covariate and the choice of the intermediate quantile level. Our findings provide practical guidelines for selecting appropriate methods and highlight the strengths and limitations of each approach.

Open via your library →

Cite this paper

https://doi.org/https://doi.org/10.1111/insr.70023

Or copy a formatted citation

@article{laurent2026,
  title        = {{Extreme Conditional Quantile Estimation in High Dimensions: A Comparative Study}},
  author       = {Laurent Gardès & Alex Podgorny},
  journal      = {International Statistical Review},
  year         = {2026},
  doi          = {https://doi.org/https://doi.org/10.1111/insr.70023},
}

Paste directly into BibTeX, Zotero, or your reference manager.

Flag this paper

Extreme Conditional Quantile Estimation in High Dimensions: A Comparative Study

Flags are reviewed by the Arbiter methodology team within 5 business days.


Evidence weight

0.50

Balanced mode · F 0.40 / M 0.15 / V 0.05 / R 0.40

F · citation impact0.50 × 0.4 = 0.20
M · momentum0.50 × 0.15 = 0.07
V · venue signal0.50 × 0.05 = 0.03
R · text relevance †0.50 × 0.4 = 0.20

† Text relevance is estimated at 0.50 on the detail page — for your query’s actual relevance score, open this paper from a search result.