|
Organizers |
A Correction of Jaccard Similarity Index For Chance Agreement In Cluster Analysis
by
Ahmed N. Albatineh
Nova Southeastern University, Fort Lauderdale, Florida
One distinction of Jaccard similarity index is that its expectation under fixed marginal totals can not be found easily because it is not linear in the matching counts, hence it can not be corrected for chance agreement. In this talk an approximation of Jaccard's expectation will be presented, and hence a correction for chance agreement can be computed. Simulation results before and after correction for a null and structured data will be presented using data generated from a mixture of bivariate normal distributions. Simulations show that the proposed approximation and hence correction for chance agreement improves the performance of the index.
Date received: July 11, 2007
Copyright © 2007 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # caur-56.