Activity Number:
|
167
|
Type:
|
Topic Contributed
|
Date/Time:
|
Monday, August 1, 2016 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistical Learning and Data Science
|
Abstract #320864
|
|
Title:
|
Studying the Importance of Variables for Clustering
|
Author(s):
|
Volodymyr Melnykov* and Yana Melnykov and Xuwen Zhu
|
Companies:
|
University of Alabama and University of Alabama and University of Alabama
|
Keywords:
|
finite mixture model ;
model-based clustering ;
variable selection ;
classification ;
EM algorithm
|
Abstract:
|
It is well documented in literature that the choice of variables severely affects the results of cluster analysis and classification. This happens due to the fact that some variables carry important clustering information while others contain redundant or no clustering information. Several approaches to identifying variables important for clustering are proposed in recent years. We discuss a novel approach to the problem, with illustrative examples and experimental justification.
|
Authors who are presenting talks have a * after their name.