INVESTIGATION OF DIFFERENT END PARTITIONS OBTAINED DURING AGGREGATIVE NONHIERARCHIC CLUSTER-ANALYSIS OF GLASSES

被引:4
作者
BARTEL, HG
NOFZ, M
WIHSMANN, FG
机构
[1] Department of Chemistry of the Humboldt-University, 1080 Berlin
[2] Central Institute of Inorganic Chemistry, the Academy of Sciences, the GDR, 1199 Berlin
关键词
D O I
10.1016/0169-7439(90)80043-6
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
The influence of number of clusters and initial partition on the result of a partitioning cluster analysis is discussed. This discussion leads to the development of an algorithm to make full use of the different end partitions found and to construct a generalized partition. The application of this algorithm is demonstrated using the physical properties of glasses as an example of objects which are to be classified. © 1990.
引用
收藏
页码:73 / 78
页数:6
相关论文
共 4 条
[1]  
Spath, Muller, Das Austauschverfahren für die skalierungsinvariante Methode der adaptiven Distanzen in der Cluster-Analyse, Ausgewhählte Operations Research Software in FORTRAN, pp. 143-163, (1979)
[2]  
Spath, Cluster-Analyse-Algorithmen zur Objektklassifizierung und Datenreduktion, (1975)
[3]  
Steinhausen, Langer, Clusteranalyse. Einführung in Methoden und Verfahren der automatischen Klassifikation, (1977)
[4]  
Nofz, Marek, Wihsmann, Bartel, Methoden der statistischen Datenanalyse zur Beschreibung von Beziehungen zwischen Zusammensetzung und Eigenschaftsdaten von Gläsern des Systems CaOAl<sub>2</sub>O<sub>3</sub>SiO<sub>2</sub>, Silikattechnik, 40, pp. 305-307, (1989)