DEPARTMENT OF MATHEMATICAL SCIENCES
University of Kentucky
The Mixturegram for Determining the Number of Components in Finite Mixture Models
ABSTRACT: In this talk, we introduce a newly-developed visualization tool for assessing the number of components in finite mixture models. This work provides a quick assessment for scientific researchers and supplements traditional numerical approaches for the problem of determining the number of components in a mixture model. An overview of the finite mixture model and the construction of the mixturegram will be presented, along with conceptual introductions to several techniques involved including EM algorithm and principal component analysis. The efficacy of this new tool is demonstrated using a simulation example and two real data applications, a univariate astronomy dataset and a multivariate cancer dataset.
Friday, January 18, 2019
2:00PM - 3:00PM
Stratton Hall 306
Pizza and soda will be served at the talk!