Table A1.

Hyperparameters optimized for each classifier and gridsearch carried out (in brackets).

RF Hyperparameters	ELM Hyperparameters	KNN Hyperparameters
Number of trees (100, 200, 500, and 750)	Number of neurons in the hidden layer (100, 500, 750, 1000, 2000, and 30,000);	Number of neighbors (1, 3, 5, and 7)
Maximum depth of these trees (6, 10, and unlimited)	Activation function (hyperbolic tangent and sigmoid).	Kernel used for weighting the distances (triangular, Biweight and Epanechnikov).
Cost of division based on the criterion of gain of information were optimized (0.001, 0.2, and 0.5)