Overview
answer as requested
1) Briefly explain how to read the tree at the bottom of the first page.
2) In a sentence, explain why the k-fold cross-validation will give a different graph every time you run it (as seen at the top of page 2).
3) Briefly compare the regression and pruned tree in terms of which you think has smaller error and which you think is more interpretable (explain your reasoning!).
3) In the error plots, the pruned tree looks like it does worse than the full tree for the 46 observations in this data set. We obviously can't use new observations in this case (there are only 46 counties in the state and we used all of them). In general which would you trust more, the full tree or the pruned tree? Give a few sentences to justify your choice.