Table 2.

Models’ parameters.

Model	Parameter	Parameter space	Chosen value
RSFs	No. trees	{25, 50, 75, 100, 250, 500, 750, 1000, 1500}	500
	Max. depth	{1, 2, 3, 4, 5, 10, 15, 25, 50, 100, 250, 500}	250
	Min. samples split	{5, 10, 15, 20, 25, 50}	15
	Min. samples leaf	{1, 2, 3, 4, 5, 10, 25}	1
SSVMs	Kernel	linear, rbf, sigmoid	rbf
	$\alpha$	Logarithmic space ranging from 0.00001 to 10	0.113
	$\gamma {}^{\ast }$	Logarithmic space ranging from 0.001 to 1	0.717
	Booster	gbtree	gbtree
XGB	No. trees	{25, 50, 75, 100, 250, 500, 750, 1000, 1500}	50
	Max. depth	{1, 2, 3, 4, 5, 10, 15, 25, 50, 100, 250, 500}	5
	Learning rate	Logarithmic space ranging from 0.01 to 1	0.05
	Subsamples	{0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8}	0.6

Each model was parametrized using a randomized search of 25 different parameter settings with a 10-fold cross validation to maximize the $c$-index.* $\gamma$ is only relevant for the rbf kernel. It was ignored for the rest.