Discuss: The values of \(y\) (the data) and \(\hat{y}\) (the model predictions) should be as close as possible.