Browse Books

Reviewer: Fatih Kurugollu

Data clustering seeks to partition data into subsets that contain data showing common properties. It is an indispensable tool in many applications, such as data analysis, pattern analysis, image processing, machine learning, and data mining. In this book, the authors examine data-clustering problems systematically. They present the existing data-clustering methods in a logical and understandable way, without going into too deep theoretical analysis. Methods are explained clearly, with enough mathematical formulation, and pseudocode for most of the algorithms is provided. In this sense, I really like how this book provides a quick reference for the clustering algorithms. I believe many researchers and engineers who need to implement a clustering algorithm will find this book very useful. However, it may not be appropriate for one who seeks a more theoretical explanation of data clustering. I do not think it can be used as a textbook in the classroom, since the book suffers from a lack of rigorous examples and questions for each chapter. However, it can be used as supplementary reading. The book is organized into four major parts, containing a total of 20 chapters. The first part is devoted to a general introduction to data clustering and related matters. Data clustering is explained in the first chapter, with required background information for the next chapters. Data types, conversions in data, standardization and transformation data before applying it to an algorithm, and the problem of visualizing the structure of data are presented in chapters 2 to 5. The last chapter of this part, chapter 6, is reserved for explanations of the similarity and dissimilarity measures that are used in the clustering algorithms. The second part, chapters 7 to 17, is the main part of the book, presenting the clustering algorithms. In each chapter, a different clustering methodology is examined, providing enough mathematical background and pseudocode for the algorithms in this category. The final chapter in this part, chapter 17, introduces the evaluation of clustering algorithms, which is of paramount importance in unsupervised clustering schemes. Since there is no prior information about the clusters, cluster shapes, and number of clusters in the data in unsupervised clustering, the result should be validated in terms of some criteria. This chapter explains the methods and criteria that are used to test the performance of clustering algorithms, as well as the ones used to estimate the number of clusters in the given data set. The third part is reserved for applications of clustering algorithms. Although the name of the part implies different applications, there is only one application: clustering gene expression data, which is considered as a case study. The background for the problem is introduced, and then a fuzzy subspace clustering algorithm is applied to the problem by providing full C++ code. The results are tabulated using a small MATLAB program, showing the closeness of data in each cluster after the clustering. In the final part of the book, some MATLAB and C++ based tools for clustering are presented. Chapter 19 offers preliminary information and techniques, as well as some built-in functions in MATLAB that are useful for clustering. Finally, in chapter 20, clustering tools in C++ are briefly presented. Online Computing Reviews Service