基于网格结构的数据流在线快速聚类算法

Online Clustering Algorithm Based on Grid Structure

针对现有的数据流聚类算法不能在线实时生成用户需要的聚类结果问题,提出一种基于滑动窗口的数据流在线聚类算法.该算法采用密度网格存储结构,实现了数据流的在线聚类过程,能实时地向用户提供聚类结果,动态地检测数据流的进化情况.实验结果表明,该方法具有快速在线聚类能力,并能保证良好的聚类质量.

As the most existing stream clustering algorithms can not generate online clustering results in real-time,an online data stream clustering algorithm is proposed by using sliding windows and density-based grid storage structure.The algorithm achieves a rapid speed for online clustering data stream and it can provide users with real-time clustering results and reflect the dynamic evolution of data streams.Experimental results show that the algorithm proposed has a good capacity of dealing with rapid evolutional data stream and have a good clustering quality.