Joint Clustering on Correlated Variables

• Activity Number: 198
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Hongmei Zhang* and Yubo Zou and Wilfried Karmaus and Graham Roberts and Hasan Arshad
• Companies: University of Memphis and BCBS and University of Memphis and University of Southampton and University of Southampton
• Keywords: Bayesian methods ; Joint clustering ; Dirichlet process

Abstract

The clustering problem has attracted much attention in the past few decades. Traditional approaches focus on the clustering of either subjects or (dependent) variables. However, clusters formed through these approaches are possibly lack of homogeneity and usually ignore the correlations between the variables. We propose a clustering method through joint clustering by clustering rows and columns considering the data is a matrix. A Bayesian method is proposed for this purpose, in which a semi-parametric model is used to evaluate any unknown relationship between variables that are possibly correlated and covariates of interest. A Dirichlet process is utilized in the process of 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 (dependent) variables and covariates, along with the ability to take the correlations between variables into account. We conduct simulation studies to examine the performance and eciency of the proposed method. The method is applied to a data set to cluster temporal sensitizations and subjects.