Variable diagnostics in model-based clustering through variation partition
Xuwen Zhu
Journal of Applied Statistics, 2018, vol. 45, issue 16, 2888-2905
Abstract:
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 solutions. In this paper, a new method through fuzzy variation decomposition is proposed for probabilistic assessing contribution of variables to a detected dataset partition. Correlation between-variable contributions reveals the underlying variable interaction structure. A visualization tool illustrates whether two variables work collaboratively or exclusively in the model. Elimination of negative-effect variables in the partition leads to better classification results. The developed technique is employed on real-life datasets with promising results.
Date: 2018
References: Add references at CitEc
Citations:
Downloads: (external link)
http://hdl.handle.net/10.1080/02664763.2018.1444740 (text/html)
Access to full text is restricted to subscribers.
Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.
Export reference: BibTeX
RIS (EndNote, ProCite, RefMan)
HTML/Text
Persistent link: https://EconPapers.repec.org/RePEc:taf:japsta:v:45:y:2018:i:16:p:2888-2905
Ordering information: This journal article can be ordered from
http://www.tandfonline.com/pricing/journal/CJAS20
DOI: 10.1080/02664763.2018.1444740
Access Statistics for this article
Journal of Applied Statistics is currently edited by Robert Aykroyd
More articles in Journal of Applied Statistics from Taylor & Francis Journals
Bibliographic data for series maintained by Chris Longhurst ().