This result indicates that a decision tree trained on one corpus of code can accurately predict coverage on a previously unseen corpus of code. The success rate of the Mockitator classifier (Figure 5(b)) was less convincing, due, once again, to Mockitator’s high ZeroR success rate.