Clustering: A Data Recovery Approach

摘要

The second edition of Mirkin's book is a refinement of Mirkin's well-received first edition. Whereas cluster analysis is often considered a set of ad hoc algorithms, Mirkin advances what he calls a "data recovery approach" to clustering. Oddly, however, the data recovery approach is not fully presented until the concluding chapter. The book focuses on K-means clustering and Ward's hierarchical clustering, but Mirkin also presents extensions and draws connections with other clustering approaches (e.g., model-based clustering, consensus clustering, etc.). An interesting feature of the book is that each chapter starts with an enumerated list of what the reader will know after reading the chapter. This is followed by "Key Concepts" giving a glossary of terms in the chapter.