Place the dataset files into this directory. For a description of the file format, see data utils.
It is recommended to place raw dataset files into subdirectories in this folder or use symlinks.
The following subdirectories are required by some parts of the code:
- clusterings: Results of clusterings applied to some datasets for further use.
- distortions: Errors (prediction - label) of some datasets used e.g. by a PredefinedDistorter.
- knn: Results of a KNN-analysis of some datasets used e.g. by the KNNPredefinedDistorter.
- subset_indices: Files containing indices of samples to split datasets into train/val/test subsets.
- unit_test: Files required for some unit tests to work.