Clustering Incomplete Data Using Normal Mixture Models

• Activity Number: 34
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Chantal Larose*+ and Dipak Dey and Ofer Harel
• Companies: University of Connecticut and University of Connecticut and University of Connecticut
• Keywords: gene expression ; missing data ; model-based clustering ; multiple imputation

Abstract

Model-based clustering using Normal mixture models provides a framework to describe how data groups together using Normal distributions. However, the existing methods for such analyses require complete data. One way to handle incomplete data is multiple imputation, a simulation-based approach which bypasses many of the disadvantages present in other methods for handling incomplete data. However, it is difficult to apply multiple imputation and cluster analysis in a straightforward manner.

In this paper, we develop a new methodology for clustering incomplete data. We have added clustering methods to particular steps in multiple imputation in order to create a way to cluster incomplete data. We illustrate how our new method outperforms existing methodology with a simulation study using Fisher's Iris dataset, then demonstrate the utility of the method on yeast gene expression data.