Table 3.

Performance of the four techniques using ANOVA and LASSO with various subsets of features

Method	Parameter value	Number of features	Model	Accuracy (%)	Sensitivity (%)	Specificity (%)	F1-score (%)
ANOVA	K = 10	10	Linear SVM	74.79	77.78	71.83	75.43
			KNN	88.67	88.63	88.71	88.62
			RF	92.09	91.77	92.42	92.03
			XGBoost	89.33	88.94	89.72	89.24
	K = 20	20	Linear SVM	74.53	77.72	71.37	75.23
			KNN	89.49	89.16	89.82	89.41
			RF	91.13	90.83	91.43	91.07
			XGBoost	89.94	89.59	90.29	89.86
	K = 30	30	Linear SVM	74.55	77.53	71.59	75.20
			KNN	90.95	90.61	91.28	90.88
			RF	91.42	91.07	91.77	91.35
			XGBoost	90.59	90.49	90.92	90.54
LASSO	C = 0.01	11	Linear SVM	74.38	77.29	71.50	75.02
			KNN	89.54	89.54	89.55	89.50
			RF	92.30	92.20	92.40	92.26
			XGBoost	89.45	89.37	89.53	89.40
	C = 0.02	21	Linear SVM	75.60	78.71	72.53	76.25
			KNN	91.23	91.16	91.29	91.19
			RF	92.29	92.15	92.43	92.25
			XGBoost	90.38	90.13	90.64	90.38
	C = 0.03	33	Linear SVM	76.02	78.80	73.26	76.58
			KNN	92.69	92.38	92.99	91.59
			RF	92.09	91.83	92.35	92.04
			XGBoost	90.83	90.69	90.96	90.77

ANOVA: analysis of variance, LASSO: least absolute shrinkage and selection operator, SVM: support vector machine, KNN: knearest neighbor, RF: random forest, XGBoost: extreme gradient boosting.