Undersampling: split into train/test and apply algorithms

Question

I am trying to build a model where there is an imbalance of classes. My steps should be:

• (1) split training/test;
• (2) apply resampling on the training