A non-parametric Bayesian model for bounded data

• We present a non-parametric Bayesian model in this paper.
• Our model has the flexibility to fit different shapes of the bounded support data.
• The number of the parameters in the proposed model is variable and infinite.
• The proposed model is tested in various data from simulated to real ones.

The intensity distribution of the observed data in many practical problems is digitalized and has bounded support. There has been growing research interest in model-based techniques to carry out on the non-Gaussian shape of observed data. However, users set remaining parameters in the existing models based on prior knowledge. Also, the distribution in the existing models is unbounded, which is not sufficiently flexible to fit different shapes of the bounded support data. In this paper, we present a non-parametric Bayesian model for modeling the probability density function of the bounded data. The advantage of our method is that the number of the parameters in the proposed model is variable and infinite, which makes the model conceptually simpler and more adaptable to the size of the data. We present numerical experiments in which we test the proposed model in various data from simulated to real data.