This paper presents a multi-density incremental clustering algorithm based on grid ( MICG) , the discriminant function taking into account relative density and gravity distance between grid cells is introduced. When a portion of the data sets changed, without re-clustering all the data, this algorithm could formulate a new cluster according to original clustering result merely based on the relationship between the unit with changed data set and neighbored unit. This approach effectively improved efficiency of cluster analysis. The time complexity and space complexity are linear with the size of dataset and the number of attributes. The experimental results show that MICG algorithm can process cluster with any shape or different densities, and can solve the incre-ment clustering problem effectively when the data is updated.%提出一种基于网格的多密度增量聚类算法MICG,定义含网格单元间的相对密度和重心距离的判别函数。当数据集的部分数据发生变动后,不需要对全部数据重新聚类,只需分析有数据变更的单元与邻居单元的关系,结合原有的聚类结果形成新的聚类,有效地提高了聚类分析的效率。时间复杂度与空间复杂度同数据集大小、属性个数成线性关系。实验结果表明,MICG算法能够处理任意形状和不同密度的类,有效地解决数据更新时的增量聚类问题。
展开▼
