KNN classification performance after using training datasets resampled with under- and oversampling methods.

Each resampling method was tested with different resampling parameters. Provided are the respective resampling parameters (other than the standard settings), resampling outputs and associated evaluation scores. As an example, Nearmiss-1 was tested with three alternative parameters which reduced the majority class from 20,635 to 722. Training on these resampled instances produced the corresponding evaluation scores on the test dataset. (xlsx)