
Industrial, Mechanical and Manufacturing Science – Zheng (Ed.)
© 2015 Taylor & Francis Group, London, ISBN: 978-1-138-02656-8
Improved K-means algorithm with better clustering centers
based on density and variance
G.C. Deng, J.C. Tao, M.J. Zhou & Y.C. Xu
School of Information Engineering, Nanchang University, Nanchang, China
ABSTRACT: Considering the defection that the traditional K-means algorithm has sensitivity to the initial
clustering centers, an improved algorithm is presented. It computes the global density of every data object, and
then adjusts the density with the variance of data object’s density area, and then finds K data objects which