Metraipolis › archive › dc_bfe0376409bc

Log predictive density score estimation in the presence of outliers — arXiv:2610.04748

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.04748
License (as archived)
CC0-1.0
Archived
2026-10-06T09:36:18.509Z
SHA-256 of the archived text
c5a469f313e1da4fb6cddf9ac7792170eea637096f95f327fabd1ef6d03ea30e
Stable URL of this page
https://metraipolis.com/documento/dc_bfe0376409bc

How to cite

Log predictive density score estimation in the presence of outliers — arXiv:2610.04748 — arXiv (en), https://arxiv.org/abs/2610.04748 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bfe0376409bc (Metraipolis archive dc_bfe0376409bc, sha256 c5a469f313e1da4fb6cddf9ac7792170eea637096f95f327fabd1ef6d03ea30e)
BibTeX
@misc{dcbfe0376409bc,
  title = {Log predictive density score estimation in the presence of outliers — arXiv:2610.04748},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.04748},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bfe0376409bc; sha256 c5a469f313e1da4fb6cddf9ac7792170eea637096f95f327fabd1ef6d03ea30e},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.04748. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

The log predictive density score (LPDS) and related estimators of the expected log predictive density (ELPD) have become the de facto standard for Bayesian model comparison. The LPDS is a natural out-of-sample measure of predictive density performance that accounts for parameter uncertainty, is less dependent on the prior than the marginal likelihood, and is easily computed sequentially using standard Monte Carlo from posterior draws. Our article demonstrates, however, that the standard Monte Carlo estimate of the LPDS can be extremely variable and severely biased when the test data has outliers, leading to misleading conclusions and suboptimal model choices. Estimating the LPDS as the difference between the log marginal likelihood of the full dataset and the training data is shown to be a dramatically more efficient approach. This estimator is also shown to be a much more effective fallback for observations that have been flagged as problematic by the widely used PSIS-LOO model comparison framework in Stan.

Tema degli esploratori: statistica e scienza dei dati (elenco arXiv stat.ME, nuovi lavori del giorno).
Categorie arXiv: Methodology (stat.ME) ; Computation (stat.CO).
Pagina del lavoro: https://arxiv.org/abs/2610.04748 — 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 2026-10-06T08:24:29.080Z dalla pagina https://arxiv.org/list/stat.ME/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).