To use either the SHAP or LIME scripts, first ensure that you have both the UNSW_NB15_testing-set.csv and UNSW_NB15_training-set.csvCSV files in your working directory. (Note: You may need to adjust the paths in the scripts to fit your device).
Then run the file based on what model need. For SHAP, the naming convention is: UNSW_SHAP_[Model].py. For LIME, the naming convention is: LIME_[Model],ipynb.
Time taken for training heavily depends on how much of the dataset is used in the Dataframe, defined by frac= parameter within the df = df.sample() function. Using a larger fraction of the dataset can lead to undesirable behaviors, including system freezing, inaccurate outputs, and out-of-memory errors.
To successfully process the entire dataset, ensure that your system is equipped with at least 64GB of RAM. Attempting to run the code with less memory WILL result in an out-of-memory error.