On Bayesian quantile regression.

Resumo: In this talk we will discuss the evolution of Bayesian quantile regression models since their first proposal and the importance of all parameters involved in the inferential process. Using a representation of the asymmetric Laplace distribution as a mixture of a normal and an exponential distribution, we discuss the importance of a scale parameter to control the variance in the model. Moreover we consider the posterior distribution of the latent variable in the mixture representation to showcase outlying observations given the Bayesian quantile regression fits, where we compare the posterior distribution of each latent variable with the others. We illustrate these results with data about Gini indexes in Brazilian states from years with census information.

Data: 03/02/2017

Horário: 11:00hs

Local : Sala 105 do PAF I.