2017-02-03

Xuwen Zhu, PhDh, Qi Zheng, PhD, Department of Mathematics, University of Louisville

"Assessment of variable contribution in model-based clustering through variation partition"

Model-based clustering is a flexible grouping technique based on fitting finite mixture models to data groups. Despite its rapid development in recent years, there is rather limited literature devoted to developing diagnostic tools for obtained clustering solution. In this paper, a new method for fuzzy variation decomposition is proposed for probabilistic assessing contribution of variables to the detected data set partition. Correlation between variable contributions reveals the underlying clustering interaction structure. Elimination of negative effect variables in the partition leads to better classification results. A visualization tool illustrates the developed technique on real-life data sets with promising results.

Stay connected TwitterFacebookLinkedInYouTubeInstagram