Laplace's method and BIC model selection for least absolute value criterion - Archive ouverte HAL Access content directly
Journal Articles Statistics and Probability Letters Year : 2022

Laplace's method and BIC model selection for least absolute value criterion

Abstract

In this paper, we provide an answer to the following question: in the particular case of a nondifferentiable likelihood, is the formula for the BIC model selection criterion the same? More precisely, we obtain the Laplace method for a sum-of-absolute-values function and we deduce that the usual BIC formula with penalty in log(n) remains the same in the context of a selection of explanatory variables by least absolute value regression.
Fichier principal
Vignette du fichier
BICL1_06072022.pdf (256.65 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03779838 , version 1 (18-09-2022)

Identifiers

Cite

Jean-Marc Bardet. Laplace's method and BIC model selection for least absolute value criterion. Statistics and Probability Letters, 2022, 195, ⟨10.1016/j.spl.2022.109764⟩. ⟨hal-03779838⟩
23 View
103 Download

Altmetric

Share

Gmail Facebook X LinkedIn More