Canopy算法中T值选取的优化及聚类效果的改进  

Optimization of T Value Selection and Improvement of Clustering Effect inCanopy Algorithm

作　　者：鲁茜 蒙祖强[1] LU Xi;MENG Zuqiang(School of computer and Electronic Information,Guangxi University,Nanning Guangxi 530004,China)

机构地区：[1]广西大学计算机与电子信息学院,广西南宁530004

出　　处：《信息与电脑》2021年第6期61-65,共5页Information & Computer

摘　　要：笔者拟通过选取距离数据样本集的中心点最近的点作为初始聚类中心,借用频率分布直方图类比出距离分布直方图,得到各数据样本点之间的距离,从而找出适宜的T1、T2取值点,实现对Canopy算法的改进。通过与K-means算法相结合,发现该改进方法能够提升算法的整体速度,同时对边缘点的聚类效果较原方法比更为清晰。利用GAUSS数据集和人工数据集对改进后的算法做聚类分析模拟实验,实验结果表明该方法在聚类效果和聚类速度上都有所提升。The author intends to select the point closest to the center point of the data sample set as the initial clustering center,borrow the frequency distribution histogram analogy to draw the distance distribution histogram,and obtain the distance between the data sample points,so as to find the appropriate T1,T2 Take the value point to realize the improvement of Canopy algorithm.By combining with the K-means algorithm,it is found that the improved method can increase the overall speed of the algorithm,and the clustering effect of edge points is clearer than the original method.Using GAUSS data set and artificial data set to do cluster analysis simulation experiments on the improved algorithm.The experimental results show that the method has improved clustering effect and clustering speed.

关 键 词：聚类算法 改进 优化 结合 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]