0

I'm working on an unbalanced dataset with three classes. The original proportions are:

  • class 1: 48%
  • class 2: 37%
  • class 3: 15%

First I'm gonna train a baseline model and I would like to know if I should resample the dataset before or after the training.

Another question is about the hyper parameters of the baseline model. What to do? Tune or not?

Thanks!

Antonio Caipora
  • 51
  • What are you trying to do? Why do you want to resample the data? – Tim Jan 06 '23 at 20:09
  • I second the questions from Tim, because the answer is that you probably shouldn’t “fix” class imbalance at all. – Dave Jan 06 '23 at 20:10
  • Actually yes, that's exactly what I was trying to do. I'm learning about machine learning and I know I cannot trust in 'accuracy' if the the dataset is unbalanced. There are other metrics I can use like 'precision', 'recall', but what if, despite Dave's advice, I still want to balance the dataset. What do you think I should do? Resample before or after train the baseline model? And about the hyper parameters of it? – Antonio Caipora Jan 07 '23 at 01:23

0 Answers0