Organization of the Book
Since the publication of the first two editions of this book, great progress has been made in the field of data mining. Many new data mining methodologies, systems, and applications have been developed, especially for handling new kinds of data, including information networks, graphs, complex structures, and data streams, as well as text, Web, multimedia, time-series, and spatiotemporal data. Such fast development and rich, new technical contents make it difficult to cover the full spectrum of the field in a single book. Instead of continuously expanding the coverage of this book, we have decided to cover the core material in sufficient scope and depth, and leave the handling of complex data types to a separate forthcoming book.
The third edition substantially revises the first two editions of the book, with numerous enhancements and a reorganization of the technical contents. The core technical material, which handles mining on general data types, is expanded and substantially enhanced. Several individual chapters for topics from the second edition (e.g., data preprocessing, frequent pattern mining, classification, and clustering) are now augmented and each split into two chapters for this new edition. For these topics, one chapter encapsulates the basic concepts and techniques while the other presents advanced concepts and methods.
Chapters from the second edition on mining complex data types (e.g., stream data, sequence data, graph-structured data, social network data, and multirelational data, as well as text, Web, multimedia, and spatiotemporal data) are now reserved for a new book that will be dedicated to advanced topics in data mining. Still, to support readers in learning such advanced topics, we have placed an electronic version of the relevant chapters from the second edition onto the book’s web site as companion material for the third edition.
The chapters of the third edition are described briefly as follows, with emphasis on the new material.
Chapter 1 provides an introduction to the multidisciplinary field of data mining. It discusses the evolutionary path of information technology, which has led to the need for data mining, and the importance of its applications. It examines the data types to be mined, including relational, transactional, and data warehouse data, as well as complex data types such as time-series, sequences, data streams, spatiotemporal data, multimedia data, text data, graphs, social networks, and Web data. The chapter presents a general classification of data mining tasks, based on the kinds of knowledge to be mined, the kinds of technologies used, and the kinds of applications that are targeted. Finally, major challenges in the field are discussed.
Chapter 2 introduces the general data features. It first discusses data objects and attribute types and then introduces typical measures for basic statistical data descriptions. It overviews data visualization techniques for various kinds of data. In addition to methods of numeric data visualization, methods for visualizing text, tags, graphs, and multidimensional data are introduced. Chapter 2 also introduces ways to measure similarity and dissimilarity for various kinds of data.
Chapter 3 introduces techniques for data preprocessing. It first introduces the concept of data quality and then discusses methods for data cleaning, data integration, data reduction, data transformation, and data discretization.
Chapters 4 and 5 provide a solid introduction to data warehouses, OLAP (online analytical processing), and data cube technology. Chapter 4 introduces the basic concepts, modeling, design architectures, and general implementations of data warehouses and OLAP, as well as the relationship between data warehousing and other data generalization methods. Chapter 5 takes an in-depth look at data cube technology, presenting a detailed study of methods of data cube computation, including Star-Cubing and high-dimensional OLAP methods. Further explorations of data cube and OLAP technologies are discussed, such as sampling cubes, ranking cubes, prediction cubes, multifeature cubes for complex analysis queries, and discovery-driven cube exploration.
Chapters 6 and 7 present methods for mining frequent patterns, associations, and correlations in large data sets. Chapter 6 introduces fundamental concepts, such as market basket analysis, with many techniques for frequent itemset mining presented in an organized way. These range from the basic Apriori algorithm and its variations to more advanced methods that improve efficiency, including the frequent pattern growth approach, frequent pattern mining with vertical data format, and mining closed and max frequent itemsets. The chapter also discusses pattern evaluation methods and introduces measures for mining correlated patterns. Chapter 7 is on advanced pattern mining methods. It discusses methods for pattern mining in multilevel and multidimensional space, mining rare and negative patterns, mining colossal patterns and high-dimensional data, constraint-based pattern mining, and mining compressed or approximate patterns. It also introduces methods for pattern exploration and application, including semantic annotation of frequent patterns.
Chapters 8 and 9 describe methods for data classification. Due to the importance and diversity of classification methods, the contents are partitioned into two chapters. Chapter 8 introduces basic concepts and methods for classification, including decision tree induction, Bayes classification, and rule-based classification. It also discusses model evaluation and selection methods and methods for improving classification accuracy, including ensemble methods and how to handle imbalanced data. Chapter 9 discusses advanced methods for classification, including Bayesian belief networks, the neural network technique of backpropagation, support vector machines, classification using frequent patterns, k-nearest-neighbor classifiers, case-based reasoning, genetic algorithms, rough set theory, and fuzzy set approaches. Additional topics include multiclass classification, semi-supervised classification, active learning, and transfer learning.
Cluster analysis forms the topic of Chapters 10 and 11. Chapter 10 introduces the basic concepts and methods for data clustering, including an overview of basic cluster analysis methods, partitioning methods, hierarchical methods, density-based methods, and grid-based methods. It also introduces methods for the evaluation of clustering. Chapter 11 discusses advanced methods for clustering, including probabilistic model-based clustering, clustering high-dimensional data, clustering graph and network data, and clustering with constraints.
Chapter 12 is dedicated to outlier detection. It introduces the basic concepts of outliers and outlier analysis and discusses various outlier detection methods from the view of degree of supervision (i.e., supervised, semi-supervised, and unsupervised methods), as well as from the view of approaches (i.e., statistical methods, proximity-based methods, clustering-based methods, and classification-based methods). It also discusses methods for mining contextual and collective outliers, and for outlier detection in high-dimensional data.
Finally, in Chapter 13, we discuss trends, applications, and research frontiers in data mining. We briefly cover mining complex data types, including mining sequence data (e.g., time series, symbolic sequences, and biological sequences), mining graphs and networks, and mining spatial, multimedia, text, and Web data. In-depth treatment of data mining methods for such data is left to a book on advanced topics in data mining, the writing of which is in progress. The chapter then moves ahead to cover other data mining methodologies, including statistical data mining, foundations of data mining, visual and audio data mining, as well as data mining applications. It discusses data mining for financial data analysis, for industries like retail and telecommunication, for use in science and engineering, and for intrusion detection and prevention. It also discusses the relationship between data mining and recommender systems. Because data mining is present in many aspects of daily life, we discuss issues regarding data mining and society, including ubiquitous and invisible data mining, as well as privacy, security, and the social impacts of data mining. We conclude our study by looking at data mining trends.
Throughout the text, italic font is used to emphasize terms that are defined, while bold font is used to highlight or summarize main ideas. Sans serif font is used for reserved words. Bold italic font is used to represent multidimensional quantities.
This book has several strong features that set it apart from other texts on data mining. It presents a very broad yet in-depth coverage of the principles of data mining. The chapters are written to be as self-contained as possible, so they may be read in order of interest by the reader. Advanced chapters offer a larger-scale view and may be considered optional for interested readers. All of the major methods of data mining are presented. The book presents important topics in data mining regarding multidimensional OLAP analysis, which is often overlooked or minimally treated in other data mining books. The book also maintains web sites with a number of online resources to aid instructors, students, and professionals in the field. These are described further in the following.