Table 5.
Hyperparameter Spaces of the Base Models and the Optimal Hyperparameters Found by Grid Search.
| Base model | Hyperparameter space | ||||
|---|---|---|---|---|---|
| Naive Bayes | var_smoothing [l,le-1,le-2,le-3,le-4,le-5,le-6,le-7,le-8,le-9] | ||||
| Linear discriminant analysis | solver [‘svd’, ‘lsqr’, ‘eigen’] | ||||
| Gaussian process | kernel [1*RBF(), l*DotProduct(), l*Matern(), l*RationalQuadratic(), 1 *WhiteKernel()] | ||||
| Support vector machine | kernel [‘rbf, ‘poly’, ‘sigmoid’, ‘linear’] | degree [1,2,3,4,5,6] | C [0.001,0.01,0.1, 1] | ||
| Decision tree | max_features [‘auto’, ‘sqrt’, ‘log2’] | ccp_alpha [0.1,0.01,0.001] | max_depth [5,6,7,8,9] | ||
| Random forest | max_.features [‘auto’, ‘sqrt’, ‘log2’] | n_estimators [1,10, 30,100, 200] | max_depth [5,6,7,8,9] | ||
| XGBoost | learning _rate [0.05,0.10,0.15] | max_depth [8, 10, 12, 15] | min_child_weight [5,7,9, 11] | gamma [0.0,0.1,0.2,0.3,0.4] | colsample_bytree [0.4, 0.5, 0.7, 1.0] |
| AdaBoost | learning _rate [0.1,1,10] | n_estimators [10,100,200] | algorithm [‘SAMME’,‘SAMME.R’] | ||
| Logistic regression | solver [ ‘lbfgs’, ‘newton-cg’, ‘liblinear’, ‘sag’, ‘saga’ ] | penalty [‘11’, ‘12’, ‘elasticnet’, ‘none’] | max_iter [100, 1000, 2000] | C [0.1,0.2,0.3,0.4,0.5] | |
| TabNet | N/A | ||||
| K-nearest neighbors | n_neighbors [1,2,3,4,5,6,7,8,9,10] | weights [‘uniform’, ‘distance’] | |||
| Multilayer perceptron | hidden_layer_sizes [(10,30,10),(10,),(10,30)] | solver [‘lbfgs’, ‘sgd’, ‘adam’] | activation [‘tanh’, ‘relu’] | learning_rate [‘constant’, adaptive’, ‘invscaling’] | alpha [0.02, 0.1, 1] |