摘要
基于网格和密度的聚类算法是一类很重要的聚类算法,但由于采用单调性搜索的方法,使得聚类结果并不十分理想,因此文中在GDD算法的基础上,提出了一种基于网格和密度的带有层次因子与距离因子的GDLD算法。GDLD算法将数据空间按要求划分成网格结构并计算网格密度,构建新的跃迁函数以达到形成有效聚类的目的。实验证明,该算法不仅能够发现任意形状的簇,而且使效率得到了很大的提高。同时层次因子既体现了簇的密度水平,也反映了簇密度的变化过程并使得算法参数更容易确定。
Clustering algorithm based on grid and density is a very important kind of clustering algorithm, but because of using monotonicity search method, it can＇ t form an effective cluster. Therefore, proposed a new clustering algorithm GDLD with hierarchy factor and distance factor based on GDD algorithm. In GDLD algorithm, data space is divided into grid structure according to the requirements and calculated the mesh density, a new transition function is constructed to form an effective cluster. Experiments show that GDLD algorithm not only can discover clusters of arbitrary shape,but also make the efficiency greatly improved. Each cluster＇ s density is reflected better and the changing process of the density is also reflected because of level factors, besides, the algorithm parameters are easier to define.
出处
《计算机技术与发展》
2012年第6期10-13,18,共5页
Computer Technology and Development
基金
天津市哲学社会科学研究规划资助项目(TJJX10-1-820)
关键词
聚类分析
层次因子
距离因子
clustering analysis
level factor
distance factor
