pySTAR Toolbox

Python Symbolic regression Through Algebraic Representations (pySTAR) provides tools for generating mathematical (symbolic) models that best fit some input data. The pySTAR toolbox allows for the definition of the form of the surrogate model and the regression parameter values simultaneously.

Reference

If you use this code please cite: "Sarwar, O. 2022, Algorithms for Interpretable High-Dimensional Regression, Carnegie Mellon University."

Installation instructions

Ensure that Anaconda is installed, and follow the steps below:

1. Open Anaconda Prompt
2. Clone the repository on your computer

git clone https://github.com/IDAES/idaes-pySTAR.git

3. Create a new conda environment called pystar by running

conda create -n pystar python==3.13
conda activate pystar

4. Navigate to the repository

cd path/to/cloned/idaes-pySTAR/repository

5. Install the package by running

pip install -r requirements-dev.txt

Configure Logger

Logger can be configured by adding the following lines at the top of your script.

from pystar import setup_logger
setup_logger()

To save the log to a file, pass the log_file argument: setup_logger(log_file="path-to-log-file.log")

Funding Acknowledgements

This work was conducted as part of the Institute for the Design of Advanced Energy Systems (IDAES) with support through the Crosscutting Research Program within the U.S. Department of Energy’s Office of Fossil Energy and Carbon Management (FECM).

Citation

If you use this code please cite: "Sarwar, O. 2022, Algorithms for Interpretable High-Dimensional Regression, Carnegie Mellon University."