Table 5.

Optimal hyperparameters of models using random search

Machine learning technique	Best feature selection method	Optimal hyperparameters values	Accuracy (%)	Sensitivity (%)	Specificity (%)	F1-score (%)
Linear SVM	Feature selection using LASSO with C = 0.03	penalty = l2 c = 92.82 verbose = false dual = false	76.03	78.81	73.27	76.59
KNN	Feature selection using LASSO with C = 0.03	n_neighbors = 1 weights = uniform algorithm = kd_tree leaf_size = 180	94.88	95.08	94.68	94.87
RF	Feature selection using LASSO with C = 0.01	n_estimators = 1000 bootstrap = true criterion = entropy max_features = none verbose = false	93.92	93.80	94.03	93.88
XGBoost	Feature selection using LASSO with C = 0.03	n_estimators = 1000 max_depth = 15 learning_rate = 0.2 objective = binary:logistic booster = gbtree gamma = 0.5 min_child_weight = 3.0 subsample = 0.8 colsample_bytree = 0.9 colsample_bylevel = 0.9 reg_alpha = 0.1 silent = false	95.31	95.19	95.43	95.28

SVM: support vector machine, KNN: k-nearest neighbor, RF: random forest, XGBoost: extreme gradient boosting, LASSO: least absolute shrinkage and selection operator.