计算机应用 ›› 2005, Vol. 25 ›› Issue (02): 294-296.DOI: 10.3724/SP.J.1087.2005.0294

• 数据库与数据挖掘 • 上一篇    下一篇

基于网格距离的聚类算法的设计、实现和应用

田启明1,2,王丽珍1,尹群1   

  1. 1.云南大学信息学院; 2.温州职业技术学院计算机系
  • 发布日期:2005-02-01 出版日期:2005-02-01
  • 基金资助:

     云南省自然科学基金资助项目(2002F0013M)

Design, realization and application of clustering algorithm based on the distance between grids

TIAN Qi-ming1,2, WANG Li-zhen1,YIN Qun1   

  1. 1. College of Information Engineering, Yunnan University, Kunming Yunan 650091, China; 2. Department of Computer Science, Wenzhou Vocational and Technological College,Zhejiang Wenzhou 325035, China
  • Online:2005-02-01 Published:2005-02-01

摘要:

提出了一种新的基于网格距离的聚类算法。该算法不仅克服了K 代表点算法中必须事先给定K值、难以确定初始代表点、聚类结果的现实意义难以描述等缺点,而且克服了基于网格的聚类算法中要求数据必须在空间密集的缺陷。通过实验验证了新算法的正确性和有效性。

关键词: 数据挖掘, 聚类, 网格, K均值聚类, 相似度量, 内涵知识

Abstract:

This paper presented a new clustering algorithm based on the distance between grids. The new algorithm not only overcomed the shortcoming of K-Medoid algorithm which has much difficulties to suppose K in advance, confirm the initialized points and explain realistically, but also overcomed the shortcoming of clustering algorithm based on grids which requests dense data in spaces. The new algorithm was proved to be correct and efficient by the results of experiments.

Key words: data mining, Clustering, grid, K-means Clustering, similarity metrics, intentional knowledge;

中图分类号: