This thesis aims to compare the ability of selected cluster analysis methods concerning classifying objects into known groups. This analysis is done by using generated quantitative data. In the real world, the “true” classification of objects into clusters is often not known. Therefore a generated type of data has been used in this thesis. The ability of the selected methods to classify objects into clusters is analysed by four external indices. Overall, the best cluster method is the Gaussian m... zobrazit celý abstraktThis thesis aims to compare the ability of selected cluster analysis methods concerning classifying objects into known groups. This analysis is done by using generated quantitative data. In the real world, the “true” classification of objects into clusters is often not known. Therefore a generated type of data has been used in this thesis. The ability of the selected methods to classify objects into clusters is analysed by four external indices. Overall, the best cluster method is the Gaussian mixture model method, followed by the hierarchical Ward method with squared Euclidean distance. Additionally, according to the indices that are analysed, the number of clusters, size of the clusters, number of variables, and different minimal cluster distance affect the ability to classify objects into clusters. |