Figure 6:
Averaged leave-one-out error obtained by the machine learning algorithm varying the number of game logs used for training.