Output details
11 - Computer Science and Informatics
University of Kent
Automating the design of data mining algorithms: an evolutionary computation approach
<24> This book proposes a sophisticated Grammar-based Genetic Programming (GGP) system that automatically creates a complete rule induction algorithm. The comprehensive evaluation of the system involved the analysis of predictive accuracy (including statistical significance) across many application domains (including bioinformatics), and several other important issues: the system's sensitivity to parameters, differences between the rule induction algorithms automatically created by the GGP system and popular manually-designed algorithms, the effect of different versions of the grammar, and the effectiveness of the GGP search versus a hill-climbing search. This work is the first to automatically create a complete data mining algorithm.
This is a 187-page, advanced research book. The technical contents of the book can be considered roughly equivalent to three journal papers. The book integrates and explains in more detail research published by the authors in two journal papers (Intelligent Data Analysis, 2009; Knowledge and Information Systems, 2009) and one chapter in an edited book (Soft Computing for Knowledge Discovery and Data Mining, 2008).