文章导读
总览 评价 康同曦 1, , 刘波 1,* , 刘强 1, , 肖燕珊 2, ( 1、 广东工业大学自动化学院; 2、 广东工业大学计算机学院; ) 摘要: 分类聚类是数据挖掘的两个主要部分,本文主要研究的是聚类算法,由于传统的聚类算法都是非此即彼的硬性聚类算法,不
康同曦1,, 刘波1,*, 刘强1,, 肖燕珊2,
(
1、广东工业大学自动化学院; 2、广东工业大学计算机学院; )
摘要:
分类聚类是数据挖掘的两个主要部分,本文主要研究的是聚类算法,由于传统的聚类算法都是非此即彼的硬性聚类算法,不能很好的满足实际生活中的一些既是A又是B的现实聚类问题。针对这个不足本文采用一种经典聚类算法,模糊函数概念的FCM算法,并在FCM算法对样本聚类效果缺陷的基础上进行改进,得到样本加权和离散系数的聚类算法,并通过Weka数据挖掘工具对经典数据集进行开发实验。通过实验证明样本加权和离散系数的聚类算法提高了样本的聚类精度降低了计算时间提高了聚类效果。
关键词:
数据挖掘;Weka; 模糊聚类; 样本加权和离散系数的聚类算法
KANG Tongxi1,, LIU Bo1,*, LIU Qiang1,, XIAO Yanshan2,
(
1、School of Automation, Guangdong University of Technology; 2、School of Computers, Guangdong University of Technology; )
Abstract:
Classification of clustering are the two main parts of data mining, this paper studies the clustering algorithm, as A result of the traditional clustering algorithm is an either/or hard clustering algorithm, not very good to meet some of the real life is both A and B clustering problem of reality. Aiming at the deficiency of a classical clustering algorithms, FCM algorithm of the concept of fuzzy functions and defects in FCM algorithm on the sample clustering effect was improved, on the basis of the weighted clustering algorithm and discrete coefficient to get samples, and through the Weka data mining tools for data sets to develop classic experiment. Experiments prove WDFCM algorithm improved the precision of the sample clustering to reduce the computation time improves the clustering effect. ?
Tag:
点此返回栏目查看更多>>>参考论文