This repository contains the code to reproduce the examples of the following paper: Numerical considerations and a new implementation for ICS, Archimbaud, A., Drmac, Z., Nordhausen, K., Radojicic, U. & Ruiz-Gazen, A. (2023). SIAM Journal on Mathematics of Data Science (SIMODS), Vol.5(1):97–121. https://epubs.siam.org/doi/10.1137/22M1498759
Two new anonymized data sets, HTP2 and HTP3 are available in the data folder in .rda format.
They are described in the paper and in the R/data.R file.
This folder is the supplementary material of the paper, containing all the plots included in the paper as well as additional ones.
It contains three R scripts:
-
all_applications.R: the script to reproduce all the results and figures from the paper. -
data.R: the description of the new data sets, HTP2 and HTP3. -
ICS_utilities.R: all self-written R functions to perform the analysis. The two main functions are:ICS_QR_full_rank()which implements the ICSQR algorithm described in the paper andICS_QR_not_full_rank()in case the data are not full rank and we want to perform first a dimension reduction based on a pivoted QR decomposition.
A file renv.lock is provided for reproducible purposes.