charcoal - Novel changepoint detection algorithms in high-dimensional linear regression

(By Fengnan Gao and Tengyao Wang)

Implementation of the changepoint localization methods via the complementary sketching algorithm (collectively named 'charcoal') for high-dimensional regression coefficients, where the regression coefficients need not be individually sparse.

• Including Algorithms 1, 2, 3 and 4 from Gao and Wang (2022).¹
• A function to generate linear regression samples with (multiple) changepoint(s) is also provided.

Description of files

• The folder singlecell contains the real data example in the paper, which came from Suo et al, 2022 and was distributed under a Creative Commons Attribution 4.0 International License.

R package

In ./R/ and ./man/ folders. Can be installed via devtools::install_github('gaofengnan/charcoal') in R.

Footnotes

1. Gao, F. and Wang, T. (2022) Sparse change detection in high-dimensional linear regression. arXiv preprint, arXiv:2208.06326. ↩