Table 5.

Summary of tuning for each model.

Model	MachineShop Model constructor	Parameters tuned	Size of tuning grid
Linear	LMModel	N/A	–
Logistic	GLMModel	N/A	–
Ridge	GLMNetModel Alpha = 0	Lambda	5
Elastic Net	GLMNetModel	Alpha, Lambda	25
LASSO	GLMNetModel Alpha = 1	Lambda	5
Neural Network	NNetModel	Size, Decay	25
SVM Poly	SVMPolyModel	C, Degree, Scale	125
SVM Radial	SVMRadialModel	C, Sigma	25
MLP	N/A	Size, Maxit, learnFuncParams	27
Random Forest	RandomForestModel	Mtry	5
GBRM	GBMModel	n.trees, interaction.depth	25

LASSO: (least absolute shrinkage and selection operator); SVM Poly: support vector machine with a polynomial kernel; SVM Radial; MLP: multi-layer perceptron; LMModel: linear model; GLMModel: generalized linear model; GLMNetModel: generalized linear model with penalized maximum likelihood; NNetModel: feed-forward neural networks; GBMModel: generalized boosted regression; Lambda: regularization parameter; Alpha: elastic net mixing parameter; Size: number of units in the hidden layer(s); Decay: parameter for weight decay; C: cost of constraints violation, regularization term in the Lagrange formulation; Degree: degree of the polynomial kernel function; Scale: scaling parameter of the polynomial kernel; Sigma: inverse kernel width; Maxit: maximum iterations to learn; learnFuncParams: parameters of the learning function; Mtry: number of variables randomly sampled as candidates at each split; n.trees: total number of trees to fit; interaction.depth: maximum depth of variable interactions.