trainstation — Convenient training of linear models

trainstation is a tool that allows the efficient construction of linear models [FraEriErh20]. It provides supplementary functionality for handling data and model metrics and relies on the scikit-learn library for the actual training [PedVarGra11]. trainstation has been originally developed to aid the construction of force constant (via hiphive [EriFraErh19]) and cluster expansions (via icet [AngMunRah19]). Since it has proven to be valuable in more general contexts and to simplify the maintenance of the two aforementioned packages, the functionality has been moved into a separate package. Many examples for the application of trainstation can be found in the documentation of hiphive and icet.

trainstation has been developed at the Department of Physics of Chalmers University of Technology in Gothenburg, Sweden. Please consult the credits page for information on how to cite trainstation.

trainstation and its development are hosted on gitlab. Bugs should be submitted via the gitlab issue tracker.

Main

• Installation
• Credits
• Module reference
  • Overview
  • Optimizer
  • EnsembleOptimizer
  • CrossValidationEstimator

Backmatter

• Bibliography
• Glossary
• Index