A bivariate Bayesian method for interval-valued regression models

As typical symbolic data, interval-valued data offer a useful tool to handle massive datasets. There has been a lot of literature focusing on researching regression models for interval-valued data based on the center and range method (CRM). However, few works are devoted to exploring Bayesian methods for interval-valued data. In this paper, we extend CRM for interval-valued regression models to the Bayesian framework for the first time. We propose a bivariate Bayesian regression model based on CRM with a known and an unknown covariance matrices, respectively. The experimental results of synthetic and real datasets show that, in contrast with classical models, the proposed Bayesian model has advantages on forecasting performances.