Table I.

Description of the Support Vector Machine algorithm hyperparameters and ranges.

Hyperparameter	Definition	Significance/usefulness	Grid search range
Kernel	Computes the dot product in the feature space using vectors from the original space.	Kernel function allows low-cost operations in the original feature space without computing the coordinates of the data in a higher dimension space.	Linear, radial basis function, polynomial, and sigmoid
γ	Distance of influence a single training point has on a kernel.	For high or low γ, training points closer or further from the decision boundary are weighted, respectively. Improved fitting of the decision boundary to training cases increases model generalizability.	0.1, 1, 10 and 100
Cost	‘C’-constant of regularization term from the Lagrange formulation (53).	Controls the trade-off between misclassifications and margin width. Simplified decision boundary for low or high cost to improve point classification.	0.1, 1, 10, 100 and 1,000