 |
Symbolic regression via genetic programming is a branch of empirical modeling that evolves summary expressions for available data. Although intrinsically difficult (the search space is infinite), recent algorithmic advances coupled with faster computers have enabled application of symbolic regression to a wide variety of industrial data sets. Unique benefits of symbolic regression include human insight and interpretability of model results, identification of key variables and variable combinations, and the generation of computationally simple models for deployment into operational models. In this presentation, we review the symbolic regression evolution process, practical issues, and approaches to managing, reviewing, and refining modeling results. A Mathematica symbolic regression package implementation will be demonstrated that stresses quality model development and a user-centric approach for model development, assessment, exploitation, and management.
|
|
For the newest resources, visit Wolfram Repositories and Archives »
