Asked about When one class is rare

Should you oversample before or after splitting?

An answer that works

After, and only the training fold. Oversampling before splitting puts copies of the same example in both train and test, which inflates the score. The validation set must keep the real class balance.

Reading an answer is not being able to give one. The section that teaches this is When one class is rare, 5 hours of Imbalanced and messy data.

Learn this properly

Also asked about Imbalanced and messy data

Your classes are 99 to 1. What do you do?How do you pick the decision threshold?Your fraud model has 0.97 ROC AUC. The fraud team says it is useless. Who is right?What does it mean for a classifier to be calibrated, and when do you need it?