Loading...
Bayesian elastic net binary quantile regression for time-dependent data analysis
Escamilla, Emilio
Escamilla, Emilio
Files
Authors
Other Names
Location
Time Period
Advisors
Original Date
Digitization Date
Issue Date
2024-05
Type
Thesis
Genre
Keywords
Subjects (LCSH)
Electronic dissertations
Citation
Abstract
Recently, various time-dependent real-life applications naturally call for the binary classification task. This predictive undertaking could potentially be challenging, especially for data collected over time. In this work, we focus on employing a binary quantile regression with autoregressive errors for the robust estimation of both the regression coefficients and autoregressive parameters. To effectively deal with larger datasets, we incorporate the elastic net penalty into our framework to simultaneously select significant covariates to enhance the interpretability and predictive powers of the final model. Studied in the Bayesian context, our binary quantile regression model is advantageous in its flexibility to employ an asymmetric Laplace working likelihood, to integrate conjugate priors on relevant parameters, to avoid the “double shrinkage problem,” to devise an efficient Metropolis-within-Gibbs sampler, and to provide a comprehensive relationship between the predictors and the output. Our proposed algorithm is illustrated through both simulation studies and real-data applications to show its efficacy and practicability when compared to some other existing methods.
Table of Contents
Description
Thesis (M.S.)-- Wichita State University, College of Liberal Arts and Sciences, Dept. of Mathematics, Statistics, and Physics
Publisher
Wichita State University
Journal
Book Title
Series
Digital Collection
Finding Aid URL
Use and Reproduction
© Copyright 2024 by Emilio Escamilla
All Rights Reserved