Modified K-Means for Better Initial Cluster Centres

The k-means clustering algorithm is most popularly used in data mining for real world applications. The efficiency and performance of the k-means algorithm is greatly affected by initial c luster centers as different initial cluster centers often lead to different clustering. In this paper, we pro pose a modified k-means algorithm which has additional steps for selecting better cluster centers. We compute Min and Max distance for every cluster and find high de nsity objects for selection of better k.

[1]  Da-Ming Zhu,et al.  Research on selecting initial points for k-means clustering , 2008, 2008 International Conference on Machine Learning and Cybernetics.

[2]  Greg Hamerly,et al.  Learning the k in k-means , 2003, NIPS.

[3]  Chen Zhang,et al.  K-means Clustering Algorithm with Improved Initial Center , 2009, 2009 Second International Workshop on Knowledge Discovery and Data Mining.

[4]  Paul S. Bradley,et al.  Refining Initial Points for K-Means Clustering , 1998, ICML.

[5]  Shijue Zheng,et al.  A New Method for Initialising the K-Means Clustering Algorithm , 2009, 2009 Second International Symposium on Knowledge Acquisition and Modeling.

[6]  Petra Perner,et al.  Data Mining - Concepts and Techniques , 2002, Künstliche Intell..

[7]  Yong Xu,et al.  K-Means Clustering Algorithm with Refined Initial Center , 2009, 2009 2nd International Conference on Biomedical Engineering and Informatics.

[8]  Fan Yang,et al.  An Improved Initialization Center Algorithm for K-Means Clustering , 2010, 2010 International Conference on Computational Intelligence and Software Engineering.