Table 10:

Validation accuracies for the random-forest models.

L	S	Fold 1	Fold 2	Fold 3	Fold 4	Average
1	2	56.23	54.77	57.97	57.21	56.55
1	3000	56.31	54.73	58.05	57.06	56.54
1	1500	56.27	54.80	58.08	56.99	56.53
1	120	56.22	54.58	58.07	57.06	56.48
1	700	56.29	54.72	57.92	56.95	56.47
1	300	56.36	54.58	57.86	56.92	56.43
1	50	56.19	54.56	57.92	56.82	56.37
5	120	56.27	54.57	57.80	56.82	56.37
5	300	56.23	54.62	57.73	56.80	56.34
1	20	56.06	54.48	57.93	56.88	56.34

We report the top 10 performing models. L denotes the minimum number of samples required to be at a leaf and S denotes the minimum number of examples required to split each internal node. We experimented with the following values: L = {1,5,20,50,150,400,800,1500}, S = {2,20,50,120,300,700,1500,3000}.