You've trained a model that looks very strong on the training set, but the team is worried it may not generalize well to new data. You want a clear way to tell whether the model is overfitting rather than simply performing well.
How would you evaluate whether a model is overfitting?