PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Discussion to Least Angle Regression
Jean-Michel Loubes and Pascal Massart
Ann. of Statistics Volume 32, Number 2, pp. 476-482, 2004.

Abstract

In this paper we discuss the Least Angle Regression algorithm proposed by Efron et al. for variable selection. In particular, in the orthogonal case we interprete their Mallows type criterion to select the number of influential variables as a penalized hard thresholding procedure. In the spirit of Birge and Massart (2004), this interpretation may lead to a data-driven strategy for penalization without knowing in advance the level of noise. Bien cordialement, Pascal Massart

EPrint Type:Article
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Computational, Information-Theoretic Learning with Statistics
ID Code:749
Deposited By:Michele Sebag
Deposited On:30 December 2004

[Edit]