Nonparametric Bayesian estimation of a bivariate density with interval censored data

Mixture of Polya trees nonparametric estimation of a bivariate density is presented for interval censored data. Real and simulated data are analyzed and compared with nonparametric maximum likelihood (NPMLE) and Bayesian G-spline estimates. An advantage of the mixture of Polya trees approach over the NPMLE is the relative ease with which continuous bivariate density and hazard plots are obtained.