Knowledge Discovery through Symbolic Regression with HeuristicLab
This contribution describes how symbolic regression can be used for knowledge discovery with the open-source software HeuristicLab. HeuristicLab includes a large set of algorithms and problems for combinatorial optimization and for regression and classification, including symbolic regression with genetic programming. It provides a rich GUI to analyze and compare algorithms and identified models. This contribution mainly focuses on specific aspects of symbolic regression that are unique to HeuristicLab, in particular, the identification of relevant variables and model simplification.
KeywordsGenetic Programming Knowledge Discovery Heuristic Optimization Symbolic Regression Tower Data
- 1.Affenzeller, M., Winkler, S., Wagner, S., Beham, A.: Genetic Algorithms and Genetic Programming - Modern Concepts and Practical Applications. Numerical Insights. CRC Press (2009)Google Scholar
- 2.Kronberger, G.: Symbolic Regression for Knowledge Discovery - Bloat, Overfitting, and Variable Interaction Networks. Trauner Verlag, Linz (2011)Google Scholar