Discriminative Information Analysis in Mixture Modeling

• Activity Number: 229
• Type: Topic Contributed
• Date/Time: Monday, August 1, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Lin Lin*+ and Cliburn Chan and Mike West
• Companies: Duke University and Duke University and Duke University
• Keywords: Bayesian expectation-maximization ; Bayesian mixture models ; Concordance of densities ; Flow cytometry data ; Multivariate mixture model ; Subset selection

Abstract

We discuss the evaluation of subsets of variables for the discriminatory evidence they provide in multivariate mixture modelling for classification. A new approach to discriminative information analysis uses a natural measure of concordance between mixture component densities. The approach is both effective and computationally attractive for routine use in assessing and prioritizing subsets of variables according to their roles in discrimination of one or more components. We relate the new discriminative information measure to misclassification rates, exemplify its use in Bayesian mixture models using novel Bayesian expectation-maximization estimation and Markov chain Monte Carlo methods, and use simulated data as an illustrative example. An application comes from the context of automatic classification and discriminatory variable selection in high-throughput systems biology using large flow cytometry data sets.