提出的增量式数据流聚类算法DGCDS结合网格和密度技术,能够得到任意形状的聚类,通过改进网格密度的计算方式,解决了现有网格算法中丢失数据空间影响信息的问题,并且实现了关键参数的自适应设置,减小了人工参数对聚类结果的影响。
This paper presented an incremental data stream clustering algorithm (DGCDS) based on grid and density, which discovered clusters with arbitrary shape. It solved a problem that losing space influence of data in some of grid-based algorithms with improving the method of density calculation, and this algorithm also set key parameter automatically to reduce the influence of factitious parameter.