|
Activity Number:
|
272
- Approaches in Clustering for Analysis of Emerging Data Types
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 9, 2022 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistical Learning and Data Science
|
|
Abstract #322872
|
|
|
Title:
|
On Measuring Soft Agreement in Clustering
|
|
Author(s):
|
Jeffrey Andrews* and Ryan Browne and Chelsey Hvingelby
|
|
Companies:
|
University of British Columbia Okanagan and University of Waterloo and Concordia University
|
|
Keywords:
|
Clustering;
Classification;
Fuzzy;
Agreement measures;
Rand index;
Mixture models
|
|
Abstract:
|
We will review some of the literature surrounding assessments of agreement between soft/fuzzy/probabilistic cluster allocations, and then provide closed-form approaches for two measures which behave as fuzzy generalizations of the popular adjusted Rand index (ARI): one novel and one previously requiring a Monte Carlo estimation process. Both proposed measures retain the reflexive property of the ARI --- in other words, that an allocation measured against itself results in the value 1.0, an arguably essential property for the interpretability of a cluster agreement measure --- and both are feasible in their closed-form for sample sizes ranging beyond five digits. We apply these methods to real and simulated data to illustrate their utility and contrast with competing approaches.
|
Authors who are presenting talks have a * after their name.
