TRANSMUTED UNIT RAYLEIGH QUANTILE REGRESSION MODEL: ALTERNATIVE TO BETA AND KUMARASWAMY QUANTILE REGRESSION MODELS

In this paper, a new alternative unit distribution is presented. It consists of applying the quadratic transmutation scheme with the unit Rayleigh distribution. The quantile regression model of the proposed distribution is developed, as well as the maximum likelihood estimation of the unknown regression model parameters. We consider a real data application that links a measure of the educational attainment of OECD (Organization for Economic Co-operation and Development) countries with some of their Better Life Index such as life satisfaction, homicide rate, and voter turnout. It is shown that the proposed quantile regression model provides a better fit than well-known regression models in the literature when the unit response variable has skewed observations and outliers.