Supplementary file for the article "Trade-off between predictive performance and FDR control for high-dimensional Gaussian model selection" - MIA-Paris Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2024

Supplementary file for the article "Trade-off between predictive performance and FDR control for high-dimensional Gaussian model selection"

Fichier supplémentaire pour l'article "Compromis entre la performance de prédiction et le contrôle du taux de fausses découvertes pour la sélection de modèles en régression linéaire gaussienne de grande dimension"

Résumé

This supplementary file is an appendix of the article entitled "Trade-off between predictive performance and FDR control for high-dimensional Gaussian model selection". This file contains three sections. Section 1 contains theoretical and empirical justifications about the estimator choice of unknown parameters $\beta^*$, $D_m^*$ and $\sigma^2$ involved in the theoretical FDR bounds in Theorem 3.2. It is a complementary work to Subsection 4.1. Section 2 contains plots as a complement to Section 1 for the bounds $B(K,\hat{\beta}_{\hat{m}(\tilde{K})},\hat{\sigma}^2)$ for each of the four scenarios described in Table 4 of Section 7. Section 3 contains studies and plots evaluating the robustness of the model collections; studies and tables of results of the random model collection constructions and results of the variable selection procedure applications for scenarios (ii), (iii) and (iv) described in Table 4 of Section 7. This last section is a complementary work to Subsections 4.3 and 4.4 of the article.
Fichier principal
Vignette du fichier
Supplementary_file_Lacroix.pdf (2.93 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04625023 , version 1 (25-06-2024)

Identifiants

  • HAL Id : hal-04625023 , version 1

Citer

Perrine Lacroix, Marie-Laure Martin. Supplementary file for the article "Trade-off between predictive performance and FDR control for high-dimensional Gaussian model selection". 2024. ⟨hal-04625023⟩
14 Consultations
4 Téléchargements

Partager

Gmail Mastodon Facebook X LinkedIn More