A Bayesian Approach for Joint Clustering Through a Dirichlet Process

• Activity Number: 123
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Yubo Zou*+ and Hongmei Zhang and Wilfried Karmaus and Hasan Arshad
• Companies: University of South Carolina and University of South Carolina and University of South Carolina and Southampton University and Hospital Trust
• Keywords: Clustering ; Bayesian method ; Gibbs sampler ; Metropolis-hastings method

Abstract

The clustering problem has attracted much attention in the past few decades. Traditional approaches focus on the clustering of either subjects or variables. However, clusters formed through these approaches are possibly lack of homogeneity. We propose a clustering method through joint clustering. Specifically, the variables are first clustered based on the agreement of relationships between variable measures and covariates, and then subjects within each variable cluster are further clustered to form refined joint clusters. A Bayesian method is proposed for this purpose, in which a semi-parametric model is used to evaluate any unknown relationship between variables and covariates, and a Dirichlet process is utilized in the process of second-step subjects clustering. The major novelty of the method exists in its ability to produce homogeneous clusters composed of a certain number of subjects sharing common features on the relationship between some variables and covariates. We conduct simulation studies to examine the performance and efficiency of the proposed method, and apply it to methylation measures at multiple CpG sites and to Zernike coefficients measures of circular pupils.