On Initial Effects of the k-Means Clustering

There are many research studies conducted in order to find a more optimal way to initialize the k-means algorithm, also referred to as Lloyd's algorithm. Despite the appreciated efficiency of the k-means process, occasionally it may return a less than optimal clustering solution. It is widely b...

Full description

Saved in:
Bibliographic Details
Published inProceedings of the International Conference on Scientific Computing (CSC) p. 200
Main Authors Burks, Sherri, Harrell, Greg, Wang, Jin
Format Conference Proceeding
LanguageEnglish
Published Athens The Steering Committee of The World Congress in Computer Science, Computer Engineering and Applied Computing (WorldComp) 01.01.2015
Online AccessGet full text

Cover

More Information
Summary:There are many research studies conducted in order to find a more optimal way to initialize the k-means algorithm, also referred to as Lloyd's algorithm. Despite the appreciated efficiency of the k-means process, occasionally it may return a less than optimal clustering solution. It is widely believed that modifications to the initialization process will improve results. Here, the choice of initial centroids for the k-means clustering technique is reviewed with respect to efficiency in stabilizing or convergence with different initialization methods. Several proposed initialization techniques are evaluated on a two dimensional model in an attempt to verify or reproduce results similar to those of the studies chosen.