A comparison of SOM neural network and hierarchical clustering methods

被引:238
作者
Mangiameli, P [1 ]
Chen, SK [1 ]
West, D [1 ]
机构
[1] E CAROLINA UNIV, SCH BUSINESS, DEPT DECIS SCI, GREENVILLE, NC 27858 USA
关键词
neural networks; cluster analysis; self organizing maps; unsupervised;
D O I
10.1016/0377-2217(96)00038-0
中图分类号
C93 [管理学];
学科分类号
12 ; 1201 ; 1202 ; 120202 ;
摘要
Cluster analysis, the determination of natural subgroups in a data set, is an important statistical methodology that is used in many contexts. A major problem with hierarchical clustering methods used today is the tendency for classification errors to occur when the empirical data departs from the ideal conditions of compact isolated clusters. Many empirical data sets have structural imperfections that confound the identification of clusters. We use a Self Organizing Map (SOM) neural network clustering methodology and demonstrate that it is superior to the hierarchical clustering methods. The performance of the neural network and seven hierarchical clustering methods is tested on 252 data sets with various levels of imperfections that include data dispersion, outliers, irrelevant variables, and nonuniform cluster densities. The superior accuracy and robustness of the neural network can improve the effectiveness of decisions and research based on clustering messy empirical data.
引用
收藏
页码:402 / 417
页数:16
相关论文
共 21 条
[11]   A METHODOLOGY FOR LARGE-SCALE RESEARCH-AND-DEVELOPMENT PLANNING BASED ON CLUSTER-ANALYSIS [J].
MATHIEU, RG ;
GIBSON, JE .
IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, 1993, 40 (03) :283-292
[12]   AN EXAMINATION OF THE EFFECT OF 6 TYPES OF ERROR PERTURBATION ON 15 CLUSTERING ALGORITHMS [J].
MILLIGAN, GW .
PSYCHOMETRIKA, 1980, 45 (03) :325-342
[13]   AN ALGORITHM FOR GENERATING ARTIFICIAL TEST CLUSTERS [J].
MILLIGAN, GW .
PSYCHOMETRIKA, 1985, 50 (01) :123-127
[14]   CLUSTER-ANALYSIS - APPLICATION OF LAGRANGIAN RELAXATION [J].
MULVEY, JM ;
CROWDER, HP .
MANAGEMENT SCIENCE, 1979, 25 (04) :329-340
[15]  
*NEUR, 1993, NEUR COMP
[16]   GENERALIZED CLUSTERING NETWORKS AND KOHONEN SELF-ORGANIZING SCHEME [J].
PAL, NR ;
BEZDEK, JC ;
TSAO, ECK .
IEEE TRANSACTIONS ON NEURAL NETWORKS, 1993, 4 (04) :549-557
[17]   CLUSTER ANALYSIS AND MATHEMATICAL PROGRAMMING [J].
RAO, MR .
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 1971, 66 (335) :622-626
[18]  
Ritter H., 1992, NEURAL COMPUTATION S
[19]  
Robles F., 1994, J DIRECT MARKETING, V8, P59
[20]  
*SAS I INC, 1988, SAS STAT US GUID REL, P283