Abstract Details
Activity Number:
|
377
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 6, 2013 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistical Learning and Data Mining
|
Abstract - #309758 |
Title:
|
Longitudinal Trajectory Cluster Analysis: How Many Groups Are There?
|
Author(s):
|
Alyssa B. Dufour*+ and L. Adrienne Cupples and Timothy Heeren and David R. Gagnon
|
Companies:
|
Hebrew SeniorLife & Harvard Medical School and Boston University and Boston University and Boston University
|
Keywords:
|
cluster analysis ;
goodness-of-fit ;
longitudinal studies ;
trajectories
|
Abstract:
|
Various methods exist for determining the correct number of groups in a cluster analysis. Studies have examined these methods using traditional cluster analysis but none have evaluated them in clustering of longitudinal data as trajectories. In this paper, we use a simulation study to examine 3 goodness-of-fit statistics used to determine how many groups exist in the data and evaluate their efficacy in cluster analysis of longitudinal trajectories using k-means methods. The statistics were Akaike's Information Criteria (AIC), Davies-Bouldin Index (DB), and Calinski-Harabasz pseudo F-statistic (CH). The latter 2 were developed specifically for determining the number of groups in a cluster analysis with a single observation per person, while the AIC is well-known for general model fit. We found that DB and CH fail to correctly identify the number of groups in the majority cases examined, while AIC was better able to determine the correct number. We also examined the number of times that all 3 criteria found the correct number of groups in the same scenario, which happened in only 8% of scenarios. The most common incorrect choice for the number of groups was choosing too few.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2013 program
|
2013 JSM Online Program Home
For information, contact jsm@amstat.org or phone (888) 231-3473.
If you have questions about the Continuing Education program, please contact the Education Department.
The views expressed here are those of the individual authors and not necessarily those of the JSM sponsors, their officers, or their staff.
Copyright © American Statistical Association.