基于相交划分的动态网格聚类算法

Dynamic grids clustering algorithm based on overlapping partition

为了解决相交网格划分技术中聚类结果对数据输入顺序的依赖性和聚类结果精度不高的问题,提出了一种基于相交划分的动态网格聚类算法(DGBO)。该算法利用相交网格划分技术和移动网格技术来解决上述问题,通过连接相交的高密度网格单元形成聚类,只需一个参数,运行速度快。实验表明,DGBO算法能够快速有效地对任意形状、大小的数据集进行聚类,并能很好地识别出孤立点和噪声。

In order to solve the problems that data input reliance and the clustering result was not efficient in overlapping grids partition technique,this paper proposed a dynamic grids clustering algorithm based on overlapping partition（DGBO）.In DGBO,used overlapping grids partition technique and shifting grids technique for solving above problems.With the help of a density threshold,merge overlapping dense grids to form clusters.As the experimental results show that DGBO algorithm can discover clusters in noisy datasets containing clusters of arbitrary shapes and sizes effectively and efficiently.