$4.$ The training error of a decision tree with ONE leaf node is $10/40$ and $40$ is the total number of training samples. What is the pessimistic error of this decision tree: $?$ After the splitting of the leaf node, there are four leaf nodes, and the training error becomes $9/40.$ Then what is the pessimistic error of the new decision tree: $?$ Should you prune the sub$-$tree after the splitting $($yes$/$no$)$: $?$