What stands out most in the summary tree is the failure of linear models which only improve 2.5 % of the time. Furthermore, the fact that the tree did not branch out from this subnode means that no further filtering of this data would have resulted in a positive result.