Banca de QUALIFICAÇÃO: PEDRO HENRIQUE ANTONACI GAMA

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : PEDRO HENRIQUE ANTONACI GAMA
DATE: 01/07/2026
TIME: 14:30
LOCAL: Ambiente virtual
TITLE:

Evaluation of selection criteria in GJS regression with variable dispersion: a simulation
study.

KEY WORDS:

GJS regression, model selection criteria, variable dispersion, Monte Carlo
simulation.

PAGES: 30
BIG AREA: Ciências Exatas e da Terra
AREA: Probabilidade e Estatística
SUMMARY:

Model selection in regression consists of choosing a relevant subset of covariates
from a wider set of candidate regressors, aiming for parsimonious models with a good quality
of fit. This work aims to investigate the performance of different model selection criteria in the
GJS (Generalized Johnson SB) regression, proposed by Lemonte and Bazán (2016). The study
adapts the methodology of Bayer (2011) — originally developed for beta regression with
variable dispersion — to the context of GJS regression, in scenarios with different degrees and
strategies of identifiability of the median and dispersion submodels. The evaluation is
conducted through Monte Carlo simulations, considering finite sample sizes and different
generating distributions of the GJS class. The best-performing selection strategy is also applied
to real data, verifying the fit of the selected model through graphical tools for residual analysis.

COMMITTEE MEMBERS:
Presidente - 2310429 - ARTUR JOSE LEMONTE
Interno - 3010614 - ELIARDO GUIMARAES DA COSTA
Interno - 1342138 - RODRIGO MATHEUS ROCHA DE MEDEIROS