This book is an excellent reference for pattern recognition, machine learning, and data mining. It focuses on the problems of classification and clustering, the two most important general problems in these areas. This book has tremendous breadth and depth in its coverage of these topics; it is clearly the best book available on the topic today.
The new edition is an excellent up-to-date revision of the book. I have especially enjoyed the new coverage provided in several topics, including new viewpoints on Support Vector Machines, and the complete in-depth coverage of new clustering methods.
This is a standout characteristic of this book: the coverage of the topics is solid, deep, and principled throughout. The book is very successful in bringing out the important points in each technique, while containing lots of interesting examples to explain complicated concepts. I believe the section on dimensionality reduction is an excellent exposition on this topic, among the best available, and this is just one example. Combined with a coverage unique in its extend, this makes the book appropriate for use as a reference, as a textbook for upper level undergraduate or graduate classes, and for the practitioner that wants to apply these techniques in practice.
I am a professor in Computer Science. Although pattern recognition is not my main focus, I work in the related fields of data mining and databases. I have used this book for my own research and, very successfully, as teaching material. I would strongly recommend this book to both the academic student and the professional.