Bayesian analysis of left censored regression with normal-compound gamma priors

This paper presents a Bayesian estimation of left censored regression models with scale mixture of normal-compound gamma priors. We presented a new hierarchical modeling for Bayesian inference in left censored regression models. We derived a Gibbs sampling algorithm from this Bayesian hierarchical modeling to estimate the regression parameters with an efficient EM algorithm for updating the hyperparameters. We illustrated the new model using simulation studies and a real data analysis. The results show that the proposed model performs very well in comparison to the other existing models.