A New Three-Parameter Flexible Unit Distribution and Its Quantile Regression Model

Abstract

This paper introduces a novel Poisson-unit-Weibull (PUW) distribution, which is defined on a unit domain and characterized by three parameters. The PUW distribution is capable of accommodating diverse non-monotone failure rates. The paper explores several significant statistical properties of the model, including the explicit closed-form expressions for the $r^{th}$ moments, quantile function, and Shannon entropy. The parameters of the PUW distribution are estimated using maximum likelihood estimation (MLE) and Bayes estimation with a square error loss function. The performance of these estimation methods is evaluated through Monte Carlo simulation studies. Furthermore, the paper discusses the practical aspects of the PUW-quantile regression model and its MLE, employing residual analysis in simulation studies. The flexibility of the PUW and PUW-quantile regression model is demonstrated through six real-life applications, showcasing their superior performance when compared to other popularly used models.

Keywords

• Unit-Weibull
• moments
• entropy
• quantile regression
• residual analysis
• maximum likelihood