基于特征选择的混合属性数据聚类挖掘仿真  

Simulation of Mixed Attribute Data Clustering Mining Based on Feature Selection

作　　者：程海军[1] 廖志雄[1] 王士斌[2] CHENG Hai-jun;LIAO Zhi-xiong;WANG Shi-bin(Xinke College,Henan Institute of Science and Technology,Xinxiang Henan 453000,China;College of Computer and Information Engineering,Henan Normal University,Xinxiang Henan 453000,China)

机构地区：[1]河南科技学院新科学院,河南新乡453000 [2]河南师范大学计算机与信息工程学院,河南新乡453000

出　　处：《计算机仿真》2020年第7期399-403,共5页Computer Simulation

基　　金：河南省自然基金面上项目(F2016038)。

摘　　要：传统算法对混合属性数据做聚类处理时,会出现大量不重要特征,从而影响数据聚类质量,使挖掘结果准确率和聚类纯度下降,造成聚类效果较差。因此提出基于特征选择的混合属性数据聚类挖掘算法,对在线聚类和离线聚类进行分析,采用数值型和分类型两种数据度量方法构建混合属性数据聚类挖掘模型,在此基础上,对数据进行预处理,通过数据紧致性和离散性加强聚类的质量,根据特征选择数据聚类算法更新聚类中心,移除冗余特征,且迅速找到最优子集,实现混合属性数据聚类挖掘。实验结果表明,上述算法能够提高混合属性数据聚类挖掘的准确率和聚类纯度,具有较好的聚类效果,可以广泛应用在现实生活中。The traditional algorithm may lead to a large number of unimportant features,affecting the clustering quality of data.Therefore,a mixed attribute data clustering mining algorithm based on feature selection was proposed.Firstly,the online clustering and offline clustering were analyzed.Then,the mixed attribute data clustering mining model was constructed by numerical and classified data measurement methods.On this basis,the data were preprocessed,and the quality of clustering was enhanced through data compactness and discreteness.Moreover,the clustering center was updated by feature selection data clustering algorithm,and the redundant features were removed.Finally,the optimal subset was found quickly.Thus,the mixed attribute data clustering mining was achieved.Simulation results show that the proposed algorithm can improve the accuracy of clustering mining of mixed attribute data and the clustering purity,so it has good clustering effect,which can be widely applied in real life.

关 键 词：特征选择 混合属性 数据聚类 冗余特征 最优子集 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]