Metraipolis › archive › dc_be822429cf64

Just add noise: Debiasing tree-based variable importance in mixed data — arXiv:2609.14083

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.14083
License (as archived)
CC0-1.0
Archived
2026-10-16T00:15:21.724Z
SHA-256 of the archived text
986cb8241989b912061b63577d5c4e0d1060ae288f46b31c8c389af1b107f8ef
Stable URL of this page
https://metraipolis.com/documento/dc_be822429cf64

How to cite

Just add noise: Debiasing tree-based variable importance in mixed data — arXiv:2609.14083 — arXiv (en), https://arxiv.org/abs/2609.14083 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_be822429cf64 (Metraipolis archive dc_be822429cf64, sha256 986cb8241989b912061b63577d5c4e0d1060ae288f46b31c8c389af1b107f8ef)
BibTeX
@misc{dcbe822429cf64,
  title = {Just add noise: Debiasing tree-based variable importance in mixed data — arXiv:2609.14083},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.14083},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_be822429cf64; sha256 986cb8241989b912061b63577d5c4e0d1060ae288f46b31c8c389af1b107f8ef},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.14083. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Variable importance scores from tree-based methods such as random forests favor continuous predictors over categorical ones. We present a theoretical analysis of this bias and propose a simple remedy: add a small amount of noise to each categorical predictor. The correction is demonstrated on a variety of simulated and real-world datasets and combined with integrated path stability selection to perform variable selection with false discovery control for mixed data.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-15 in cs.LG).
Categorie arXiv: Machine Learning (stat.ML) ; Machine Learning (cs.LG); Methodology (stat.ME).
Pagina del lavoro: https://arxiv.org/abs/2609.14083 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-15T23:54:14.881Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).