No label defined (Q4040865)

An introduction to methods of cluster analysis is given with emphasis on applications in medicine. Most of basic notions from cluster analysis are formulated in terms of graph theory. The problem of finding an appropriate decomposition of a given basic set with respect to a measure of dissimilarity is representated as a problem of finding components of connectivity in a special graph or multigraph. The first three parts of the book can be read without any previous knowledge from cluster analysis or graph theory. These parts will be useful for nonmathematicians who want to use methods of cluster analysis. In parts 4 and 5 the problem of validity of decomposition of a basic set given by proposed cluster methods is discussed. The probabilistic models based on random graphs and multigraphs and their relation to the validity problem are investigated. These parts would be interesting also for specialists in graph theory. The last part of the book is devoted to three complete examples of the previous methods.