2 points How does a classication tree for a twoclass response variable decide which variable (and value) to use for a particular decision (split)? Which of the following is a true statement about the splitting in tree classication? Select the variable (and value) that most decreases the difference in class quantities in each of the child nodes relative to the parent node. Select the variable (and value) that most increases the difference in class proportions in each of the child nodes relative to the parent node. Select the variable (and value) that most increases the difference in class quantities in the parent node relative to each of the child nodes. Select the variable (and value) that most increases the difference in class quantities in each of the child nodes relative to the parent node. Select the variable (and value) that most decreases the difference in class proportions in each of the child nodes relative to the parent node. 2 points Q Consider bootstrap aggregation, or bagging, of classication trees. Which of the following statements about bagging in this context is the only one that is not true? The predicted class for a particular data point based on the bagged trees is the class with the highest proportion of trees predicting that class. The structures of the trees for each bootstrap sample tend to be very similar, although predicted classications from each tree can vary widely' Bagging ts a classication tree to each of a collection of bootstrap samples of the training data and nds the proportion of trees predicting each class at each data point. It is possible to gauge the importance of each predictor when bagging classification trees. a 2 points Consider bootstrap aggregation, or bagging, of a regression tree. Which of the following statements about bagging in this context is true? Bagging tends to decrease mean squared prediction error by reducing bias while keeping variance unchanged. Bagging tends to increase mean squared prediction error by reducing bias while keeping variance unchanged. Bagging tends to decrease mean squared prediction error by reducing variance while keeping bias unchanged. Bagging tends to increase mean squared prediction error by reducing variance while keeping bias unchanged