Table 4. Random Forest (RF), Gradient Boosting (GB), and AdaBoost (AB) methods: Hyperparameters’ domain and the corresponding tuned values at the data sets under consideration.

The N_e, M_ss, M_sl, lr, in respect, represents the number of estimators, minimum number of samples per split, minimum number of samples per leaf, and learning rate.

method / data set	parameters
	N e	M ss	M sl	lr
AB	{10, 11, …, 10000}	-	-	[1e-3, 5e-1]
RF	{10, 11, …, 10000}	{2, 3, …, 10}	{1, 2, …, 10}	-
GB	{10, 11, …, 10000}	{2, 3, …, 10}	{1, 2, …, 10}	[1e-3, 5e-1]
AB at Demo	545	-	-	0.017
RF at Demo	6197	5	9	-
GB at Demo	257	4	1	0.005
AB at Fixation	415	-	-	0.169
RF at Fixation	2726	2	10	-
GB at Fixation	3380	3	6	0.007
AB at IA	4736	-	-	0.087
RF at IA	1980	4	3	-
GB at IA	165	2	10	0.160
AB at Demo-Fixation	309	-	-	0.215
RF at Demo-Fixation	9923	9	1	-
GB at Demo-Fixation	2674	9	3	0.282
AB at Demo-IA	7133	-	-	0.019
RF at Demo-IA	163	2	1	-
GB at Demo-IA	971	7	3	0.299