CLUSTERING ALGORITHM THROUGH THE COMMON LINEARITY TREND AMONG CLUSTERS AND ITS APPLICATION

DOI Web Site 5 References
  • Kuroki Manabu
    Department of Systems Innovation, Graduate School of Engineering Science, Osaka University
  • Hirata Masaru
    Nagoya city office
  • Miyakawa Masami
    Department of Industrial Engineering and Management, Graduate School of Decision Science and Technology, Tokyo Institute of Technology

Bibliographic Information

Other Title
  • 群間に共通な線形傾向を仮定したクラスタリング法とその応用
  • グン カン ニ キョウツウ ナ センケイ ケイコウ オ カテイ シタ クラスタリングホウ ト ソノ オウヨウ

Search this article

Description

The aim of the k-means method is to divide n samples into k clusters such that the within-cluster sum of squares is minimized. However, the clusters generated by the k-means method show only similarity of points within clusters. In addition, it is difficult to visualize the results by k-means method when the number of variables is greater than two. In this paper, in order to characterize a relationship between clusters, k-means method is extended to k-planes method as the new non-hierarchical clustering method. By using this method, we can generate such clusters that show not only similar points within each cluster but also common linear trend among clusters. In addition, the k-means method can be regarded as the zero-dimensional k-planes method. Furthermore, the results of k-planes method can be visualized through the projection on the complementary space of the given hyperplane.

Journal

References(5)*help

See more

Related Projects

See more

Keywords

Details 詳細情報について

Report a problem

Back to top