Mathematical classification and clustering (Q2564098)

To write a book about classification and clustering is not an easy task. There is the fact that the whole world is organized through clusters: scientific knowledge is based on concepts which are ideal clusters; laws are applicable to groups of people satisfying certain conditions, i.e. clusters; social life is organized in classes or clusters. With these facts in mind the author gives an overall analysis and view of classification and cluster theory. He starts with an enumeration and definitions of the different forms of classifications, followed by the types of data. The central part of the book presents and classifies all the different forms of algorithms, useful for someone looking for the best technique to solve a particular problem. The three main techniques, `subset', `partition' and `hierarchy', are studied in depth in three different chapters. Here, all aspects of the methods are analysed and supplemented with useful statements and demonstrations. Many numerical examples illustrate the behaviour of the algorithms, proper to each method. The different chapters are well written, with adequate subdivisions for the different aspects of the subjects. Clearly visible frames surround the most important statements that the author wants to stress and many subdivisions close with a few discussion points. This makes the book pleasant and efficient to read. One regret: only little attention is paid to most recently developed classes of techniques such as fuzzy clustering and scattering matrix based methods. Overall it should be emphasized that this book is a monumental work with a large treasure of information. It certainly contains one of the most complete compilation of classification and clustering methods together with all relevant technical and theoretical explanations. The author should be praised for this most interesting work.