Consistency of modified versions of Bayesian Information Criterion in sparse linear regression with subgaussian errors

2014/11/15 by Piotr Szulc, Szulc, Piotr
Engineering · Mathematics · #62J05 #Advanced Statistical Methods and Models #FOS: Mathematics #Sparse and Compressive Sensing Techniques #Statistical Methods and Inference #Statistics Theory (math.ST)

paper · pdf · doi:10.48550/arxiv.1411.4138

openalex publication_date 2014/11/15 · openalex created_date 2018/06/13 · openalex updated_date 2026/07/28

Abstract

We consider a sparse linear regression model, when the number of available predictors, p, is much larger than the sample size, n, and the number of non-zero coefficients, p₀, is small. To choose the regression model in this situation, we cannot use classical model selection criteria. In recent years, special methods have been proposed to deal with this type of problem, for example modified versions of Bayesian Information Criterion, like mBIC or mBIC2. It was shown that these criteria are consistent under the assumption that both n and p as well as p₀ tend to infinity and the error term is normally distributed. In this article we prove the consistency of mBIC and mBIC2 under the assumption that the error term is a subgaussian random variable.

Citations

• Asymptotic Bayes optimality under sparsity for generally distributed effect sizes under the alternative
• Combined l₁ and greedy l₀ penalized least squares for linear model selection

Related

• Bayesian model choice and information criteria in sparse generalized linear models
• On the overestimation of widely applicable Bayesian information criterion
• Unified Bayesian theory of sparse linear regression with nuisance parameters
• Semiparametric Bayesian Information Criterion for Model Selection in Ultra-high Dimensional Additive Models
• Bayesian High-dimensional Semi-parametric Inference beyond sub-Gaussian Errors