Inferential and predictive procedures for inverse gamma regression model

The maximum likelihood method is often applied by practitioners in a statistical analysis relying on the assumption inferences can be drawn based on asymptotic properties. However, the analysis can be compromised when the sample size is small. In this work, we developed a skewness coefficient considering it as a criterion to if or not the inference procedure is reliable when the inverse gamma regression model is the candidate to fit a dataset. Additionally, a prediction interval was further developed for this model. An extensive simulation study and two captivating databases to, respectively, evaluate and illustrate the methodologies demonstrated here are included in this work.