Review of "Data Mining: Practical Machine Learning Tools and Techniques" by Witten and Frank

In the early 1990s some sectors of the computer science community were developing the idea of data understanding as a discovery-driven, systematic and iterative process. This "data mining" research and development area was expected to take advantage of the expansion and consolidation of machine learning methodologies together with the integration of traditional statistical analysis and database management strategies. The main goal was to identify relevant, interesting and potentially novel informational patterns and relationships in large data sets to support decision making and knowledge discovery. In the mid 1990s developers and users of decision-making support systems in areas such as finance (e.g. credit approval and fraud detection applications), marketing and sales analysis (e.g. shopping patterns and sales prediction) were showing a great deal of enthusiasm about the business value of data mining applications. During the next few years international conferences, journals and books were more frequently reporting advances, tools and applications in other areas such as biomedical informatics, engineering, physics, law enforcement and agriculture. Today data mining is seen as a discipline or paradigm that actively aids in the development of these and other scientific areas (e.g. Web-based computing and systems biology). Data mining has become a fundamental research topic in the progression of computing applications in health care and biomedicine. Advances in data mining have applications and implications in areas ranging from information management in healthcare organisations, consumer health informatics, public health and epidemiology, patient care and monitoring systems, large-scale image analysis to information extraction and classification of scientific literature [1]. Approaches, techniques and applications associated with data mining has also significantly supported different data understanding and decision support tasks in bio-signal processing, such as the classification, visualisation and identification of complex relationships between diagnostic variables or groups of patients [2,3].