Flexible Robust Mixture Regression Modeling

Marcus G. Lavagnole Nascimento; Carlos A. Abanto-Valle

doi:10.57805/revstat.v20i1.365

Flexible Robust Mixture Regression Modeling

Authors

Marcus G. Lavagnole Nascimento Universidade Federal do Rio de Janeiro https://orcid.org/0000-0002-9742-5683
Carlos A. Abanto-Valle Universidade Federal do Rio de Janeiro https://orcid.org/0000-0003-0862-4152

DOI:

https://doi.org/10.57805/revstat.v20i1.365

Keywords:

Finite mixture of regressions, scale mixture of skew-normal distributions, Markov chain Monte Carlo

Abstract

This paper provides a flexible methodology for the class of finite mixture of regressions with scale mixture of skew-normal errors (SMSN-FMRM) introduced by [42], relaxing the constraints imposed by the authors during the estimation process. Based on the data augmentation principle and Markov chain Monte Carlo (MCMC) algorithms, a Bayesian inference procedure is developed. A simulation study is implemented in order to understand the possible effects caused by the restrictions and an example with a well known dataset illustrates the performance of the proposed methods.

Downloads

REVSTAT2022v20n1_6

Published

2022-02-01

How to Cite

Lavagnole Nascimento , M. G., & Abanto-Valle , C. A. (2022). Flexible Robust Mixture Regression Modeling. REVSTAT-Statistical Journal, 20(1), 101–115. https://doi.org/10.57805/revstat.v20i1.365