Lasso penalized model selection criteria for high-dimensional multivariate linear regression analysis

Shota Katayama, Shinpei Imori

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

This paper proposes two model selection criteria for identifying relevant predictors in the high-dimensional multivariate linear regression analysis. The proposed criteria are based on a Lasso type penalized likelihood function to allow the high-dimensionality. Under the asymptotic framework that the dimension of multiple responses goes to infinity while the maximum size of candidate models has smaller order of the sample size, it is shown that the proposed criteria have the model selection consistency, that is, they can asymptotically pick out the true model. Simulation studies show that the proposed criteria outperform existing criteria when the dimension of multiple responses is large.

Original languageEnglish
Pages (from-to)138-150
Number of pages13
JournalJournal of Multivariate Analysis
Volume132
DOIs
Publication statusPublished - 2014 Nov 1
Externally publishedYes

Fingerprint

Model Selection Criteria
Lasso
Linear regression
Regression Analysis
Regression analysis
High-dimensional
Multiple Responses
Penalized Likelihood
Likelihood Function
Model Selection
Dimensionality
Predictors
Sample Size
Infinity
Simulation Study
Model selection criteria
Model

Keywords

  • Consistency
  • High-dimensional data
  • Model selection
  • Multivariate linear regression

ASJC Scopus subject areas

  • Statistics and Probability
  • Numerical Analysis
  • Statistics, Probability and Uncertainty

Cite this

Lasso penalized model selection criteria for high-dimensional multivariate linear regression analysis. / Katayama, Shota; Imori, Shinpei.

In: Journal of Multivariate Analysis, Vol. 132, 01.11.2014, p. 138-150.

Research output: Contribution to journalArticle

@article{82a0ec71e22f472083f9d0cc356ba43a,
title = "Lasso penalized model selection criteria for high-dimensional multivariate linear regression analysis",
abstract = "This paper proposes two model selection criteria for identifying relevant predictors in the high-dimensional multivariate linear regression analysis. The proposed criteria are based on a Lasso type penalized likelihood function to allow the high-dimensionality. Under the asymptotic framework that the dimension of multiple responses goes to infinity while the maximum size of candidate models has smaller order of the sample size, it is shown that the proposed criteria have the model selection consistency, that is, they can asymptotically pick out the true model. Simulation studies show that the proposed criteria outperform existing criteria when the dimension of multiple responses is large.",
keywords = "Consistency, High-dimensional data, Model selection, Multivariate linear regression",
author = "Shota Katayama and Shinpei Imori",
year = "2014",
month = "11",
day = "1",
doi = "10.1016/j.jmva.2014.08.002",
language = "English",
volume = "132",
pages = "138--150",
journal = "Journal of Multivariate Analysis",
issn = "0047-259X",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - Lasso penalized model selection criteria for high-dimensional multivariate linear regression analysis

AU - Katayama, Shota

AU - Imori, Shinpei

PY - 2014/11/1

Y1 - 2014/11/1

N2 - This paper proposes two model selection criteria for identifying relevant predictors in the high-dimensional multivariate linear regression analysis. The proposed criteria are based on a Lasso type penalized likelihood function to allow the high-dimensionality. Under the asymptotic framework that the dimension of multiple responses goes to infinity while the maximum size of candidate models has smaller order of the sample size, it is shown that the proposed criteria have the model selection consistency, that is, they can asymptotically pick out the true model. Simulation studies show that the proposed criteria outperform existing criteria when the dimension of multiple responses is large.

AB - This paper proposes two model selection criteria for identifying relevant predictors in the high-dimensional multivariate linear regression analysis. The proposed criteria are based on a Lasso type penalized likelihood function to allow the high-dimensionality. Under the asymptotic framework that the dimension of multiple responses goes to infinity while the maximum size of candidate models has smaller order of the sample size, it is shown that the proposed criteria have the model selection consistency, that is, they can asymptotically pick out the true model. Simulation studies show that the proposed criteria outperform existing criteria when the dimension of multiple responses is large.

KW - Consistency

KW - High-dimensional data

KW - Model selection

KW - Multivariate linear regression

UR - http://www.scopus.com/inward/record.url?scp=84908554860&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84908554860&partnerID=8YFLogxK

U2 - 10.1016/j.jmva.2014.08.002

DO - 10.1016/j.jmva.2014.08.002

M3 - Article

AN - SCOPUS:84908554860

VL - 132

SP - 138

EP - 150

JO - Journal of Multivariate Analysis

JF - Journal of Multivariate Analysis

SN - 0047-259X

ER -