Quantile Regression for positive data using a general class of distributions

TL;DR

This paper introduces a new class of quantile regression models for positive data based on the IRON distribution, along with inference tools, diagnostics, and an R package, demonstrated through household income analysis.

Contribution

It proposes a novel IRON-based quantile regression framework for positive data, with accompanying inference methods and an R package for practical application.

Findings

The IRON-based models effectively fit positive continuous data.

The R package IRON facilitates estimation and diagnostics.

Application to Chilean household income demonstrates practical utility.

Abstract

This paper presents a general class of quantile regression models for positive continuous data. In this class of models we consider that the response variable has a IRON distribution. We provide inference and diagnostic tools for this class of models. An R package, called IRON, was implemented. This package provides estimation and inference for the parameters and tools useful to check the fit of models. The methods are also illustrated with an application to modeling household income in Chile.