A new Bayesian approach to robustness against outliers in linear regression

DOI10.1214/19-BA1157MaRDI QIDQ2226686zbMATH OpenOpenAlexWikidataFDO

Authors Philippe Gagnon, Alain Desgagné, M. Bédard

Publication date 9 February 2021

Published in Bayesian Analysis (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1612.06198, https://projecteuclid.org/euclid.ba/1558598428

zbMATH Keywords

variable selectionANOVAmaximum likelihood estimationANCOVAbuilt-in robustnesssuper heavy-tailed distributionswhole robustness

Mathematics Subject Classification ID

Nonparametric robustness (62G35) Linear regression; mixed models (62J05) Extreme value theory; extremal stochastic processes (60G70) Statistics of extreme values; tail inference (62G32) Analysis of variance and covariance (ANOVA) (62J10)

Abstract: Linear regression is ubiquitous in statistical analysis. It is well understood that conflicting sources of information may contaminate the inference when the classical normality of errors is assumed. The contamination caused by the light normal tails follows from an undesirable effect: the posterior concentrates in an area in between the different sources with a large enough scaling to incorporate them all. The theory of conflict resolution in Bayesian statistics (O'Hagan and Pericchi (2012)) recommends to address this problem by limiting the impact of outliers to obtain conclusions consistent with the bulk of the data. In this paper, we propose a model with super heavy-tailed errors to achieve this. We prove that it is wholly robust, meaning that the impact of outliers gradually vanishes as they move further and further away form the general trend. The super heavy-tailed density is similar to the normal outside of the tails, which gives rise to an efficient estimation procedure. In addition, estimates are easily computed. This is highlighted via a detailed user guide, where all steps are explained through a simulated case study. The performance is shown using simulation. All required code is given.

Recommendations

• Bayesian robustness to outliers in linear regression and ratio estimation
• Bayesian likelihood robustness in linear models
• scientific article; zbMATH DE number 3905641
• Robust Bayesian Regression Analysis Using Ramsay-Novick Distributed Errors with Student-t Prior
• scientific article; zbMATH DE number 1843026

Cites work

• scientific article; zbMATH DE number 3905641 (Why is no real title available?)
• scientific article; zbMATH DE number 3905646 (Why is no real title available?)
• scientific article; zbMATH DE number 3986407 (Why is no real title available?)
• A Bayesian approach to some outlier problems
• A Useful Convergence Theorem for Probability Distributions
• A class of robust and fully efficient regression estimators
• Asymptotic behavior of Bayes estimates under possibly incorrect models
• Bayesian heavy-tailed models and conflict resolution: a review
• Bayesian likelihood robustness in linear models
• Bayesian robustness modelling of location and scale parameters
• Bayesian robustness to outliers in linear regression and ratio estimation
• Equation of state calculations by fast computing machines
• Full robustness in Bayesian modelling of a scale parameter
• High breakdown-point and high efficiency robust estimates for regression
• Monte Carlo sampling methods using Markov chains and their applications
• Rejection of Outliers
• Reversible jump Markov chain Monte Carlo computation and Bayesian model determination
• Robust linear regression: A review and comparison
• Robust regression: Asymptotics, conjectures and Monte Carlo
• Robustness to outliers in location-scale parameter model using log-regularly varying distributions
• Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images

Cited in

(16)

• Robustness against conflicting prior information in regression
• Log-regularly varying scale mixture of normals for robust regression
• Robust heavy-tailed versions of generalized linear models with applications in actuarial science
• Posterior robustness with milder conditions: contamination models revisited
• Bayesian robustness to outliers in linear regression and ratio estimation
• Short proof of posterior robustness: An illustration of basic ideas in a simple case
• On the robustness to outliers of the Student‐t process
• Bayesian likelihood robustness in linear models
• Bayesian modal regression based on mixture distributions
• scientific article; zbMATH DE number 1843026 (Why is no real title available?)
• Theoretical properties of Bayesian Student-\(t\) linear regression
• Efficient and robust estimation of regression and scale parameters, with outlier detection
• Robustness against outliers: A new variance inflated regression model for proportions
• A principal component analysis of multivariate data on inflation for Nigeria
• scientific article; zbMATH DE number 3905641 (Why is no real title available?)
• Informed reversible jump algorithms

This page was built for publication: A new Bayesian approach to robustness against outliers in linear regression

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2226686)