Traditional data mining concentrates on extracting patterns from existing datasets. While very useful, pattern extraction provides relatively little guidance for businesses when it comes to decision making for the future.

This concise (88 page) book introduces readers to the basic concepts of proactive data mining with decision trees. The book is a part of the “Springer Briefs in Electrical and Computer Engineering” series.

The authors propose using a decision tree model to proactively classify datasets while mining through existing ones. The model is not only able to predict and explain a phenomenon in the datasets; it also utilizes a problem’s domain knowledge to suggest specific actions for achieving optimal changes in the values of the target attributes. The model works in two phases. In the first phase, it trains a probabilistic classifier using the datasets with a supervised learning algorithm. The second phase uses the classification model produced in the training phase to select potential actions to maximize the utility while reducing the costs. In such selection, the model allows the users to examine different results by varying the input parameters to an extent.

The book is divided into six chapters. Chapter 1 introduces readers to the basic concepts behind proactive data mining and explains why the approach is more practical and useful for business. The argument is that the decision tree approach can lead to results that make more intuitive sense and that can help businesses make decisions. Chapter 2 outlines a general approach for proactive data mining and its algorithmic framework. The key elements here are to allow variations in the input data and to use the utility function to assess the changes in the results due to different inputs and different utility functions. Chapter 3 describes one specific approach of proactive data mining, the approach that uses decision trees. The utility functions and the optimizing methods are then implemented in the form of decision trees. Chapter 4 provides two case studies with some data collected from real-world applications, one from a cellular service system and the other from a security company. Chapter 5 discusses the sensitivity analysis to see what changes would come from variations of data and utility functions, followed by a summary of the book in chapter 6.

The book is very well written, easy to understand, and easy to follow. Each chapter is well organized. The chapters in the book are well connected. Abundant references are provided at the end of each chapter. Anyone who has some data mining and artificial intelligence background will be able to get a clear picture from this book about what proactive data mining with decision trees is. The book is especially useful for practitioners who would like to get started in using data mining tools for business applications.