Symulacyjna optymalizacja wyboru procedury klasyfikacyjnej dla danego typu danych – oprogramowanie komputerowe i wyniki badań, s. 120-129

Abstract

In typical cluster analysis study eight major steps are distinguished (see Milligan [1996, 342-343]; Walesiak [2005]). Four of them represent the critical steps: decisions concerning variable normalisation formula, selection of a distance measure, selection of clustering method, determining the number of clusters. The article presents: a) determination of optimal clustering procedure for a data set by varying all combinations of normalization formulas, distance measures, and clustering methods. Nine paths of simulation was separated depends on variable scale of measurement in a data set; b) clusterSim computer program written in R and C++ languages; c) some empirical results of simulation study based on data matrix with growing number of objects and variables.