首页 | 本学科首页   官方微博 | 高级检索  
     

数据流的不规则网格增量聚类算法
引用本文:于翔,印桂生.数据流的不规则网格增量聚类算法[J].哈尔滨工程大学学报,2008,29(8).
作者姓名:于翔  印桂生
作者单位:哈尔滨工程大学计算机科学与技术学院,黑龙江哈尔滨,150001
摘    要:分析了数据流的特点,针对数据流聚类算法CluStream对数据流中非球形聚类效果不好的情况,提出了基于数据流的不规则网格增量聚类算法IIGStream.IIGStream算法具备了传统网格聚类算法处理速度快的优点.同时能够动态增量地调整网格结构.对新到来的数据点,通过判断网格是否相连,保证了对于不同形状聚类的聚类效果.IIGStream在聚类时无需预先指定聚类数目.且对孤立点不敏感.在真实数据集与仿真数据集上的实验结果表明,IIGStream算法具有良好的适用性和有效性,在聚类精度以及速度上均优于CluStream算法.

关 键 词:数据流  聚类  网格  数据挖掘

An incremental irregular grid algorithm for clustering data streams
YU Xiang,YIN Gui-sheng.An incremental irregular grid algorithm for clustering data streams[J].Journal of Harbin Engineering University,2008,29(8).
Authors:YU Xiang  YIN Gui-sheng
Abstract:CluStream algorithm,as a data stream clustering algorithm,has a poor effect in treatment of non-spherical shape clustering.So an incremental clustering algorithm for irregular grid,named IIGStream,was developed to solve the problem.IIGStream adjusts the grid structure dynamically and incrementally while keeping the advantage of fast clustering which traditional grid clustering algorithms have.For new data points,by judging whether the grid is interconnected,it ensures the clustering effect for various shapes.In the process of clustering,it need not specify the number of clusters in advance,and it is not sensitive to outliers.Experiments on both real datasets and simulation datasets proved the applicability and validity of IIGStream and it outperformed CluStream in both precision and speed.
Keywords:data stream  clustering  grid  data mining
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号