Knowledge discovery in databases. Techniques and applications (Q4494739)

This textbook is an introduction to the methods of data mining or Knowledge Discovery in Databases (KDD). The foundations of KDD are dealt with focusing on data base systems (relational model, SQL, physical storage and data structures), descriptive and inductive statistics and basic probability theory (such as random variables, approximation of distributions, parameter estimation and hypothesis testing). The basic techniques and algorithms of data mining and its applications are discussed in single chapters. These contain a wide range of statistical clustering techniques such as \(k\)-means and single link clustering, subspace and increment clustering, as well as approaches to efficient clustering on very large databases. The chapter on classification systems deals with Bayesian classifiers, nearest-neighbor classifiers and decision trees, including their extension to large data sets. The search for simple association rules (using the a priori algorithm) is continuously refined to incorporate taxonomies (leading to hierarchic association rules) and numerical attributes (yielding quantitative association rules). The notion of generalization from single data is dealt with from the perspective of the manual interactive notion of online analytical processing based on data cubes, as well as automatic generalization by attribute-oriented induction. One chapter deals with complex types of data to be considered, which lead to temporal data mining, spatial data mining, text and Web mining. The book concludes with a brief consideration of alternative automatic discovery methods such as inductive logic programming, genetic algorithms, neural networks, and Kohonen maps.