Why is it generally a good idea to prune a full regression tree?

a. A full regression tree is too large to visualize.

b. A full regression tree typically overfits the training data and pruning the tree improves the predictive performance on validation data.

c. Pruning a tree by applying restrictions on how much a variable splitting rule can divide up a node of observations decreases the training set variance of the outcome variable in the resulting leaf nodes.

d. Pruning a tree results in a weak, unstable learner that is better for out-of-sample predictions.