This is a python script to conduct analysis on the impact of synthetic resampling techniques on the performance of binary classification models. For more information please check out our publication in EJNMMI.
To avoid clashes in package dependancies, we strongly recommend using Anaconda 3 enviroments which can be downloaded in from their website. All requirements are given in requirements.txt.
For example from the anaconda3 prompt:
(base) C:\> cd path-to-this-directory
(base) C:\path-to-this-directory> conda create -n resample python=3.6
(base) C:\path-to-this-directory> conda activate resample
(resample) C:\path-to-this-directory> pip install -r requirements.txt
For usage of the script please use the help argument:
python SynResampleClass.py --help
We provided two example datasets in the examples\ directory. It is important to set the --target and --index headers as the same as the header in the data csv. For example:
python SynResampleClass.py examples\TextureSession_DFS_v2.csv result_output --target DFS --index ID
If there are any problems or questions, please email du94@hku.hk