Bayesian Variable Selection in Clustering via Dirichlet Process Mixture Models

 

Variable selection has been the focus of much research in recent years. This talk will focus on the the development of Bayesian methods for variable selection in problems that aim at clustering the samples. A novel methodology will be described that uses infinite mixture models via Dirichlet process mixtures to define the cluster structure. A latent binary vector identifies the discriminating variables and is updated via a Metropolis algorithm. Inference on the cluster structure is obtained via a split-merge MCMC technique. Performances of the methodology are illustrated on simulated data and on DNA microarray data.