A heuristic method for finding the optimal number of clusters with application in medical data

Hamidreza Bayati; Heydar Davoudi; Emad Fatemizadeh

doi:10.1109/IEMBS.2008.4650258

A heuristic method for finding the optimal number of clusters with application in medical data

Annu Int Conf IEEE Eng Med Biol Soc. 2008:2008:4684-7. doi: 10.1109/IEMBS.2008.4650258.

Authors

Hamidreza Bayati¹, Heydar Davoudi, Emad Fatemizadeh

Affiliation

¹ Biological Signal and Image Processing Laboratory (BiSIPL), Department of Electrical Engineering, Sharif University of Technology, Iran.

PMID: 19163761
DOI: 10.1109/IEMBS.2008.4650258

Abstract

In this paper, a heuristic method for determining the optimal number of clusters is proposed. Four clustering algorithms, namely K-means, Growing Neural Gas, Simulated Annealing based technique, and Fuzzy C-means in conjunction with three well known cluster validity indices, namely Davies-Bouldin index, Calinski-Harabasz index, Maulik-Bandyopadhyay index, in addition to the proposed index are used. Our simulations evaluate capability of mentioned indices in some artificial and medical datasets.

MeSH terms

Algorithms
Artificial Intelligence
Cluster Analysis*
Computer Simulation
Data Interpretation, Statistical
Fuzzy Logic
Gene Expression Profiling
Humans
Medical Informatics / methods*
Models, Genetic
Models, Statistical
Models, Theoretical
Pattern Recognition, Automated / methods