L2E Package for R


An R package for implementing a computational framework for L2E structured regression problems


About the package

The L2E package (version 2.0) implements the computational framework for L2E regression in Liu, Chi, and Lange (2022+), which was built on the previous work in Chi and Chi (2022). Both works employ the block coordinate descent strategy to solve a nonconvex optimization problem but utilize different methods for the inner block descent updates. We refer to the method in Liu, Chi, and Lange (2022+) as “MM” and the one in Chi and Chi (2022) as “PG” in our package. This package provides code to replicate some examples illustrating the usage of the frameworks in both manuscripts.

Demo code

For demo code showing how to use the framework, please refer to the GitHub page for the package at https://jocelynchi.github.io/L2E-package-demo/.

Citing L2E

To cite the L2E framework, please use the following BibTeX entries.

@article{l2e,
  author = {Jocelyn T. Chi and Eric C. Chi},
  title = {A User-Friendly Computational Framework for Robust Structured Regression with the L$_2$ Criterion},
  journal = {Journal of Computational and Graphical Statistics},
  year = {2022},
  doi = {10.1080/10618600.2022.2035232}
}
@article{L2E-Liu,
  title={A Sharper Computational Tool for L$_2$E  Regression},
  author={Liu, Xiaoqian and Chi, Eric C. and Lange, Kenneth},
  journal={arXiv preprint arXiv:2203.02993},
  year={2022}
}