Efficiency assessment of Ichino method and mean value method of selecting variables in cluster analysis Cover Image

Ocena efektywności metody uśredniania zmiennych i metody Ichino selekcji zmiennych w analizie skupień
Efficiency assessment of Ichino method and mean value method of selecting variables in cluster analysis

Author(s): Jerzy Korzeniewski
Subject(s): Economy
Published by: Wydawnictwo Uniwersytetu Ekonomicznego we Wrocławiu
Keywords: cluster analysis; variable selection; mutual neighbourhood graph; taking mean values of variables

Summary/Abstract: Ichino method can be applied to variables measured on any type of scale. Pełka and Wilk examined the method but their investigation was limited to a couple of cluster structures and the criterion used was the stability of grouping based on the silhouette index. The mean value method [Fraiman et al. 2008] is a relatively new method not investigated until now in a broad simulation experiment. In the paper the results of simulation experiment are presented. The criteria used are: recall, precision and Rand index. The competing methods are HINoV and its two modifications: VSKM and VAF with clusterability index.

  • Issue Year: 2012
  • Issue No: 242
  • Page Range: 444-450
  • Page Count: 7
  • Language: Polish