|
Activity Number:
|
460
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
| Abstract - #305717 |
|
Title:
|
Model-Based Clustering in a Brook Trout Classification Study within the Eastern United States
|
|
Author(s):
|
Huizi Zhang*+ and Samantha C. Prins and Eric P. Smith
|
|
Companies:
|
Virginia Polytechnic Institute and State University and Virginia Polytechnic Institute and State University and Virginia Polytechnic Institute and State University
|
|
Address:
|
Department of Statistics, 1216 University City Blvd., Blacksburg, VA, 24060,
|
|
Keywords:
|
model-based clustering ; Voronoi tessellations ; stressor-response ; classification ; brook trout
|
|
Abstract:
|
Cluster analysis is a commonly used technique on multivariate data that aims to group objects into clusters such that objects are similar within each cluster and dissimilar between different clusters. We developed a model-based clustering method for analysis of ecological data that groups objects by their empirical stressor-response relationship, rather than their attribute values, using Voronoi tessellations to find the optimal grouping solution. In particular, we extended the Voronoi approach to deal with categorical response data. This required the development of appropriate optimality criteria. This method is applied to a carefully compiled dataset of Brook trout presence/absence within the eastern United States. Results indicate fairly strong relationships that vary spatially.
|
