Table 2.

Hyperparameters for grid search.

Algorithms and parameter name			Search space		Optimal
Logistic regression
	Penalty C tol solver multi_class	(‘l1’, ‘l2’, ‘none’) (0.01, 0.1, 1.0) (0.0001, 0.001, 0.01) (‘lbfgs’, ‘liblinear’, ‘sag’, ‘saga’) (‘auto’, ‘ovr’, ‘multinomial’)		l2 1.0 0.0001 lbfgs auto
Random forest
	n_estimators max_depth max_features min_samples_splitmin_samples_leaf	(5, 10, 50, 100, 150) (1, 2, 3, 5, None) (’auto’, ’sqrt’) (2, 5, 10) (1, 2, 4)		100 None auto 2 1
Extra trees
	n_estimators max_depth max_features min_samples_splitmin_samples_leaf	(5, 10, 50, 100, 150) (1, 2, 3, 5, None) (’auto’, ’sqrt’) (2, 5, 10) (1, 2, 4)		100 None auto 2 1
Gradient boosting trees
	Loss n_estimators max_depth learning_rate criterion	(‘deviance’, ‘exponential’) (5, 10, 50, 100, 150) (1, 2, 3, 5) (0.001, 0.01, 0.1) (’friedman_mse’, ’mse’, ’mae’)		deviance 100 3 0.1 friedman_mse