Table 2.

Classifier optimized hyperparameters and variation range.

Classifier	Hyperparameter	Variation range
k-nearest neighbour (k-NN)	Distance (DD)	{Cityblock, chebychev, correlation, cosine, euclidean, hamming, jaccard, mahalanobis, minkowski, seuclidean, spearman}
	DistanceWeight (DW)	{Equal, inverse, squaredinverse}
	Exponent (E)	[0.5, 3]
	NumNeighbors (NN)	[1, 5]
Support Vector Machine (SVM)	BoxConstraint (BC)	Log-scaled in the range [1e−3, 1e3]
	KernelFunction (KF)	{Gaussian, linear, polynomial}
	KernelScale (KS)	Log-scaled in the range [1e−3, 1e3]
	PolynomialOrder (PO)	{1,2,3,4}
Artificial Neural Network (ANN)	Activation Function (AF)	{relu, sigmoid, tanh}
	Hidden Layer nr. of Neurons (HLN)	[25, 200]
Linear Discriminant Analysis (LDA)	Gamma (G)	[0,1]
	Delta (D)	Log-scaled in the range [1e−6, 1e3]
	DiscrimType (DT)	{Linear, quadratic, diagLinear}
		{diagQuadratic, pseudoLinear, pseudoQuadratic}
Naive Bayes (NB)	DistributionName (DN)	{Normal, kernel}
	Width (W)	Log-scaled in the range [1e−4, 1e14]
	Kernel (K)	{Normal, box, epanechnikov, triangle}