基于非度量多维缩放的聚类组合算法
|
摘要
针对单一聚类方法远不能满足实际数据分析需求,且K-Means聚类中维数高,非度量型数据分析亟待解决的问题,提出一种基于非度量多维缩放的聚类组合算法(NMDSCCA)。该算法通过非度量多维缩放方法对非度量型的高维数据进行降维,利用降维后得到的主成分变量作为输入变量,以K-Means算法作为基聚类器进行聚类,解决了K-Means算法无法处理分类数据以及维数高的变量局限性,使其具有普适性。仿真实验表明,新算法不仅聚类效果上均优于传统K-Means算法及基于主成分分析(PCA)的聚类组合算法,而且算法应用于大数据时具有更高的收敛速度。
Concerning the problem about real and complex data analysis not being met by single clustering method and non-metric and high-dimensional variables exited in K-Means algorithm, a Clustering Combination Algorithm based on Nonmetric MultiDimensional Scaling( NMDSCCA) was proposed. Firstly, the non-metric multi-dimensional scaling method was used to reduce the dimension. Then, using the principal component variables obtained after dimensionality reduction as input variables, and the K-Means algorithm as a base classifier for clustering, The limitations existed in K-Means algorithm about the classification of data and high-dimensional variable were solved and the algorithm was made universal. The simulation results show that the algorithm not only has advantages over both traditional K-Means algorithm and clustering algorithm based on Principal Component Analysis( PCA) in cluster performance experiments, but also has high convergence speed when dealing with big data.
Concerning the problem about real and complex data analysis not being met by single clustering method and non-metric and high-dimensional variables exited in K-Means algorithm, a Clustering Combination Algorithm based on Nonmetric MultiDimensional Scaling( NMDSCCA) was proposed. Firstly, the non-metric multi-dimensional scaling method was used to reduce the dimension. Then, using the principal component variables obtained after dimensionality reduction as input variables, and the K-Means algorithm as a base classifier for clustering, The limitations existed in K-Means algorithm about the classification of data and high-dimensional variable were solved and the algorithm was made universal. The simulation results show that the algorithm not only has advantages over both traditional K-Means algorithm and clustering algorithm based on Principal Component Analysis( PCA) in cluster performance experiments, but also has high convergence speed when dealing with big data.
引文
[1]STREHL A, GHOSH J. Cluster ensembles:a knowledge reuse framework for combining multiple partitions[J]. Journal of Machine Learning Research, 2002, 3(3):583-617.
[2]ZHANG X L, BRODLEY C E. Solving cluster ensemble problems by bipartite graph partitioning[C]//Proceedings of the 21st International Conference on Machine Learning. New York:ACM, 2004:9-15.
[3]王敏峰,朱敏琛.一种新的聚类组合算法[J].福州大学学报(自然科学版),2010,38(6):819-823.
[4]孟子健,马江洪.一种可选初始聚类中心的改进K均值算法[J].统计决策,2014(12):12-14.
[5]韩凌波. K-均值算法个数优化问题研究[J].四川理工学院学报(自然科学版),2012,25(2):77-84.
[6]徐勇,陈亮.一种基于降维思想的K均值聚类算法[J].湖南城市学院学报(自然科学版),2017,26(1):54-61.
[7]余世孝.非度量多维测度及其在群落分类中的应用[J].植物生态学报,1995,19(2):128-136.
[8]RIVAS M N, BURTON O T, WISE P, et al. A microbiota signature associated with experimental food allergy promotes allergic sensitization and anaphylaxis[J]. Journal of Allergy and Clinical Immunologym, 2013, 131(1):201-212.
[9]王斌会.多元统计分析及R语言建模[M].广州:暨南大学出版社,2011:267-279.
[10]张学工.模式识别[M].北京:清华大学出版社,2015:173-177.
[11]周爱武,陈宝楼,王琰. K-Means算法的研究与改进[J].计算机技术与发展,2012,22(10):101-104.
[12]宋媛.聚类分析中确定最佳聚类数的若干问题研究[D].延边:延边大学,2013:11-27.
[2]ZHANG X L, BRODLEY C E. Solving cluster ensemble problems by bipartite graph partitioning[C]//Proceedings of the 21st International Conference on Machine Learning. New York:ACM, 2004:9-15.
[3]王敏峰,朱敏琛.一种新的聚类组合算法[J].福州大学学报(自然科学版),2010,38(6):819-823.
[4]孟子健,马江洪.一种可选初始聚类中心的改进K均值算法[J].统计决策,2014(12):12-14.
[5]韩凌波. K-均值算法个数优化问题研究[J].四川理工学院学报(自然科学版),2012,25(2):77-84.
[6]徐勇,陈亮.一种基于降维思想的K均值聚类算法[J].湖南城市学院学报(自然科学版),2017,26(1):54-61.
[7]余世孝.非度量多维测度及其在群落分类中的应用[J].植物生态学报,1995,19(2):128-136.
[8]RIVAS M N, BURTON O T, WISE P, et al. A microbiota signature associated with experimental food allergy promotes allergic sensitization and anaphylaxis[J]. Journal of Allergy and Clinical Immunologym, 2013, 131(1):201-212.
[9]王斌会.多元统计分析及R语言建模[M].广州:暨南大学出版社,2011:267-279.
[10]张学工.模式识别[M].北京:清华大学出版社,2015:173-177.
[11]周爱武,陈宝楼,王琰. K-Means算法的研究与改进[J].计算机技术与发展,2012,22(10):101-104.
[12]宋媛.聚类分析中确定最佳聚类数的若干问题研究[D].延边:延边大学,2013:11-27.