This might be a dumb question ! I built a model and I'm satisfied enough with the model, given that I have a dataset with categorical variables I wanted to see the R2/RMSE for each of those categories, but I'm unsure on which set I should conduct this test should I use the training set for that ? My understanding is that the results would be too good because the model already "saw" the data I'm inputing. So should I use the test set instead ?

Full article content could not be extracted automatically. Read the original below.