Bayesian Inference on Compositional Data Sampling from Dirichlet Distribution
-
Graphical Abstract
-
Abstract
A Bayesian approach to the parameters and other interesting quantities of the Dirichlet likelihood is proposed. The uniform prior is placed on the meaningful function of the parameters. After transforming the parameters, the Metropolis algorithm is used to draw the posterior samples and the results of the Bayesian inference are followed.
-
-