This webpage contains complementary material to the research paper:

	José A. Sáez and José L. Romero-Béjar. On the suitability of bagging-based ensembles with borderline label noise. Mathematics 10 (2022) 1892.

The web is organized according to the following summary:

1. Real-world datasets
2. Performance and robustness results

1. Real-world datasets

The experiment in this research is based on the following real-world datasets, which are injected with different levels of borderline label noise:

These datasets can be downloaded from the webpages:

• UCI Machine Learning Repository
• KEEL-dataset Repository

Additionally, these datasets can be downloaded here.

2. Performance and robustness results

Neighborwise borderline label noise	File
- Accuracy results per noise level
- Robustness results per noise level
- Box-plots of robustness results
Non-linearwise borderline label noise	File
- Accuracy results per noise level
- Robustness results per noise level
- Box-plots of robustness results