计算机应用 ›› 2005, Vol. 25 ›› Issue (01): 45-48.DOI: 10.3724/SP.J.1087.2005.00045

• 数据挖掘 • 上一篇    下一篇

一种带约束的最小离差平方和系统聚类法及应用

李斌1,郭剑毅2   

  1. 1.昆明理工大学生物与化学工程学院; 2.昆明理工大学信息工程与自动化学院
  • 出版日期:2005-01-01 发布日期:2005-01-01

Method and application of restricted minimum variance hierarchical cluster

LI Bin1,GUO Jian-yi2   

  1. 1. College of Biological and Chemical Engineering, Kunming University of Science and Technology; 2. College of Information Engineering and Automation, Kunming University of Science and Technology,
  • Online:2005-01-01 Published:2005-01-01

摘要: 针对异常数据的不利影响和类数难以判别的问题提出了一种带约束的最小离差平方和系统聚类法,包括对初始数据进行线性变换、聚类检验指标分析、主因素分析等。通过对实际调查样本数据进行聚类,挖掘和分析客户群中所存在的不同特征的组群,得到了直观的聚类过程和较合理的分组结果。

关键词:  数据挖掘, 聚类分析, 客户关系管理

Abstract: To the questions of adverse effect of unusual data and difficulties in defining the number of clusters, a method of Restricted Minimum Variance Hierarchical Cluster(RMVHC) was proposed, including varying initial data linearly, inspection standard analysis and main factors analysis etc. Through clustering to the sample data of investigation, excavating and analysing various characteristics existing in customers actually, cluster course and relatively rational grouping result were directly perceived through the senses.

Key words: Data Mining, clustering, customer relationship management

中图分类号: