Bayesian composite quantile regression for longitudinal count data

H. AL-Sharoot, Mohammed; Muhammed, Sabaa

doi:10.33916/qjae.2026.01106115

	Bayesian composite quantile regression for longitudinal count data
AL-Qadisiyah Journal For Administrative and Economic sciences
Volume 28, Issue 1, April 2026, Pages 106-115 PDF (704.72 K)
Document Type: Research Paper
DOI: 10.33916/qjae.2026.01106115
Authors
Mohammed H. AL-Sharoot; Sabaa Muhammed
University of AL-Qadisiya
Abstract
This paper develops a composite quantile regression framework for the analysis of longitudinal count data. Unlike classical regression approaches that assume a continuous response, the proposed method explicitly accounts for the discrete nature of count outcomes while preserving their inherent smooth structure. The model is constructed using a flexible representation based on a growing mixture of asymmetric distributions, which allows the conditional distribution of the response variable to be captured across multiple quantiles. To facilitate Bayesian inference, a structured Gibbs sampling algorithm is derived for parameter estimation. The performance of the proposed approach is carefully evaluated through extensive simulation studies, demonstrating its robustness and efficiency under various data-generating scenarios. Furthermore, the methodology is applied to a real dataset from the field of neurology, illustrating its practical relevance and interpretability. Comparative analysis with existing models highlights the advantages of the proposed composite quantile regression approach for longitudinal count data.
Keywords
Asymmetric Laplace; Gibbs sampling; longitudinal count data; Markov chain; Monte Carlo; Poisson process; quantile regression

Statistics Article View: 29 PDF Download: 64