We are doing a binary classification that classifies malicious packets. It is imbalanced data with a much smaller number of malicious packets. So we used the pr curve to select
