Because the data distribution in datasets might not be uniform, your test set might be difficult to anticipate, making it impossible to determine whether your model is inefficient