Inferência estatística via bootstrap no modelo de regressão gama unitária

Abstract

In Economics, there are many situations involving data restricted to the range (0;1), that is, data of rates and proportions, and there are models that are better suited to this situation, such as the Unit Gamma regression model . However, when the sample size is small, or even moderate, the Statistical Inference of these models is compromised. Estimators, in general, tend to become more biased and test statistics lead to less accurate tests. Thus, it is necessary to use tools that are able to correct the bias of estimators and test statistics, such as the method of bootstrap. In this work, we propose Monte Carlo simulations, via bootstrap, which solve the aforementioned problems. In addition, we study socioeconomic variables that impact energy generation through photovoltaic systems, using the Unit Gamma regression model and Statistical Inference via it bootstrap