Supplementary Files - Boulesteix (2014)

Which resampling-based error estimator for benchmark studies? A power analysis with application to PLS-LDA

by Anne-Laure Boulesteix

Supplementary files dowload:

• The data sets considered in our paper and in Feres de Souza et al (2010), that were kindly forwarded to us by Bruno Feres de Souza:datasets1.zip (84MB), datasets2.zip (49MB), datasets3.zip (41MB), datasets4.zip (31MB).
• The reproducibility files (reproducibility.zip) to reproduce all our analyses.

Instructions to reproduce the results presented in the paper:

• If you just want to reproduce the figures based on the already computed R-objects:

1. Download and unzip the ZIP-file reproducibility.zip.

2. Start R and set the working directory to the directory you have just unzipped (containing the directories results, data_txt, data_R and the files datahandling.R, analysis_comparison_paper.R, figures_comparison_paper.R). Here you only need the directory results and the file figures_comparison_paper.R

3. Run the R-code from the file figures_comparison_paper.R. The figures will appear in the directory.

• If you want to reproduce the whole analysis, i.e. to produce the R-objects based on the original txt files by yourself:

1. Download and unzip the files datasets1.zip,...,datasets4.zip.
2. Download and unzip the file reproducibility.zip.
3. Start R and set the working directory to the directory you have just unzipped (containing the directories results, data_txt, data_R and the files datahandling.R, analysis_comparison_paper.R, figures_comparison_paper.R).
4. In the resulting directory 'reproducibility', there is an empty directory called 'data_txt'. Cut and paste all the data txt files from the directories datasets1,...,datasets4 into this empty directory data_txt.
5. Run the R-code from the file datahandling.R to import and transform the data sets from the directory data_txt into R-objects and store them in the directory data_R.
6. Run the R-code from the file analysis_comparison_paper.R to generate the results and store them in the directory results.