ReadMe for the repository associated with:

"Interpretation of High-Dimensional Linear Regression: Effects of Nullspace and Regularization Demonstrated on Battery Data"

https://arxiv.org/abs/2309.00564

Author: Joachim Schaeffer
Email: [email protected]

Code

All plots associated with the paper can be generated by the notebooks. The code for objects, methods, and functions is in the src folder.

Example of fully synthetic parabolic data:

• Nullspace_Parabola_Examples.ipynb

Example of Lithium-Iron-Phosphate Cycling data:

• Nullspace_LFP_examples.ipynb

LFP Data Files

The measurement data is contained in the file lfp_slim.csv, and the corresponding license for this data is lfp_datalicense.txt.

Environments

All this Python code and tooling has dependencies which are encoded in the environment files. To install the anaconda environment, you need to have anaconda installed, then run:

conda env create --file python_environments/environment.yml
conda activate HDRegAnalytics

Known issue: You need to have a working installation of latex and all other requirements for matplotlib to work with latex. More information here MatplotlibLatex.

We recommend using R-Studio for running the R code contained in the folder regression_in_R.

License

The code is licensed according to the terms of the AGPL-3.0 License. The license for the LFP data is included in the data folde:

• lfpdatalicense.md

Acknowledgment/Citation

If you use code, results, or ideas from this repository for your work, please cite the following:

"Interpretation of High-Dimensional Linear Regression: Effects of Nullspace and Regularization Demonstrated on Battery Data"

@misc{schaeffer2023interpretation,
title={Interpretation of High-Dimensional Linear Regression: Effects of Nullspace and Regularization Demonstrated on Battery Data},
author={Joachim Schaeffer and Eric Lenz and William C. Chueh and Martin Z. Bazant and Rolf Findeisen and Richard D. Braatz},
year={2023},
eprint={2309.00564},
archivePrefix={arXiv},
primaryClass={stat.ML}
}