Finally suppose you apply boyh models from $($d$)$ to a data set with $2,000$ test observatiins. What fraction of tje time do you expect the predicted class labels from your model to agree with those from your friend's model. Explain your answer.