A Regression Model for Positive Data Based on the Slashed Half-Normal Distribution

Abstract

In this paper, we discuss several aspects about the slashed half-normal distribution. We reparameterize the model based on the mean and we perform comparisons with well-known regression models for positive data. Maximum likelihood estimation of the parameters is carried out through the expectation-maximization algorithm. Some properties of the estimators and two kinds of residuals are assessed in a simulation study. Two real datasets illustrated the proposed model as well as other three models for the sake of comparison.