|
Activity Number:
|
383
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #306490 |
|
Title:
|
Estimating the Number of Data Clusters via Agreement Measure--Based Statistics
|
|
Author(s):
|
Heng Liu*+ and Michelle Wang and Douglas Simpson
|
|
Companies:
|
University of Illinois at Urbana-Champaign and University of Illinois at Urbana-Champaign and University of Illinois at Urbana-Champaign
|
|
Address:
|
Illini Hall, Champagin, IL, 61820,
|
|
Keywords:
|
kappa statistics ; fMRI time series ; clustering ; naive Bayes rule
|
|
Abstract:
|
Clustering and classification have been important tools to address a broad range of problems in fields such as image analysis, genomics, and other areas. Many different heuristic clustering criteria are available. Often the data partitioning suffers lack of consistency across different algorithms and criteria. In this paper we propose a Kappa-type of agreement measure to compare and combine results from different clustering methods, and to select the number of clusters. By maximizing the agreement on allocation of the observations between different clustering methods, the approach is intended to provide a robust consensus set of clusters. The favorable performance of the method is demonstrated in simulation studies and for real fMRI time series data. Finally the asymptotic properties of the Kappa statistics are discussed.
|
