Automating the design of data mining algorithms. An evolutionary computation approach (Q1041166)

Traditionally, data mining is seen as the non-trivial extraction of implicit, previously unknown and potentially useful information from data. The techniques and corresponding algorithms used in data mining are borrowed from diverse domains, such as statistics, machine learning, pattern recognition, etc. Classically, when using data mining models, the researcher has to manually design his (her) own algorithms or apply others, after a previous modification, to match the concrete problem to solve. This book tries to avoid this situation, proposing an alternative consisting in an automatic generation of new data mining algorithms. Basically, a potential user will simply ask the computer to automatically generate a new algorithm, specially designed for his (her) data. This represents an interesting approach in using data mining algorithms, and the comparison between the traditional way and this new one will certainly strengthen both of them. Technically, the main contributions of the book are presented in Chapter 5, where the authors describe a new genetic programming system to automate the design of rule induction algorithms, and in Chapter 6, focused on reporting the results of computational experiments evaluating the new system. The book will be useful for postgraduate students and researchers in the data mining field and in evolutionary computation.