DPP: Inference of Parameters of Normal Distributions from a Mixture of Normals
Luis M. Avila, Michael R. May, Jeff Ross-Ibarra · 2017
This MCMC method takes a data numeric vector (Y) and assigns the elements of Y to a (potentially infinite) number of normal distributions. The individual normal distributions from a mixture of normals can be inferred. Following the method described in Escobar (1994) we use a Dirichlet Process Prior (DPP) to describe stochastically our prior assumptions about the dimensionality of the data.