Bayesian Model Selection for Misspecified Models in Linear Regression

2017/06/11 by MB de Kock, de Kock, MB, HC Eggers +1
Computer Science · Mathematics · #62-07 #62F15 #62J05 #Bayesian Methods and Mixture Models #Blind Source Separation Techniques #FOS: Computer and information sciences #Methodology (stat.ME) #Statistical Methods and Inference

paper · pdf · doi:10.48550/arxiv.1706.03343

openalex publication_date 2017/06/11 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

While the Bayesian Information Criterion (BIC) and Akaike Information Criterion (AIC) are powerful tools for model selection in linear regression, they are built on different prior assumptions and thereby apply to different data generation scenarios. We show that in the finite-dimensional case their respective assumptions can be unified within an augmented model-plus-noise space and construct a prior in this space which inherits the beneficial properties of both AIC and BIC. This allows us to adapt the BIC to be robust against misspecified models where the signal to noise ratio is low.

Citations

• Model selection for ecologists: the worldviews of AIC and BIC

Related

• Multimodel Inference
• AIC and BIC
• Semiparametric Bayesian Information Criterion for Model Selection in Ultra-high Dimensional Additive Models
• Bayesian Information Criterion for Linear Mixed-effects Models
• Tractable Bayesian variable selection: beyond normality