Automating the Design of Data Mining Algorithms

Más información sobre el libro

Data mining is a vibrant research field with numerous real-world applications, utilizing concepts and methods to extract valuable knowledge from datasets, aiding decision-making across various sectors. Despite the availability of numerous data mining algorithms, selecting the most suitable one for specific problems remains challenging. Additionally, existing algorithms are often manually designed, reflecting human biases and preferences. This book introduces a novel approach to algorithm design by advocating for the systematic automation of data mining algorithm creation through evolutionary computation. Specifically, it focuses on a genetic programming system, a method that evolves computer programs, to automate the design of rule induction algorithms, which are essential for discovering classification rules from data. The emphasis on genetic programming is due to its effectiveness in automating program generation and its capability for global search within the solution space of data mining algorithms. The book also acknowledges the potential for exploring other search methods for this task in the future.