摘要
CFSFDP聚类算法适应于任意形状的类簇,不需要提前设定聚类数,通过对局部密度和距离的计算产生决策图,从而人工选择聚类中心.若聚类中心在决策图中基本重叠时,肉眼无法分辨,造成对聚类中心的漏选.文章采取基于密度权值平均变化率的CFSFDP聚类算法,增加数据点之间的差异性,以偏离的变化趋势求拐点,通过计算得到聚类中心,提高聚类的准确性.
CFSFDP clustering algorithm adapts to any shape of clusters,does not need to set the number of clusters in advance,through the calculation of local density and distance to generate decision maps,so as to select the cluster center manually.If the clustering centers overlap in decision maps,they can not be distinguished by naked eyes,resulting in the omission of clustering centers.In this paper,CFSFDP clustering algorithm based on the average change rate of density weight is adopted to increase the difference between data points,and the inflection point is calculated by the trend of deviation.The clustering center is calculated to improve the stability of clustering.
作者
董炎焱
DONG Yanyan(Department of Mathematical Sciences,Jinzhong Teachers College,Jinzhong 030600,China)
出处
《太原师范学院学报（自然科学版）》
2018年第3期33-36,共4页
Journal of Taiyuan Normal University:Natural Science Edition
基金
基于职教云平台的混合式学习实践研究(GH-171547)
