Table 5.

Performance metrics of various classifiers under different feature selection methods.

	Classifier	ROC	Sensitivity	Specificity	F1 Score	Accuracy	PPV	NPV
Feature selection with Lasso	LR	0.575	0.542	0.581	0.558	0.561	0.575	0.549
	SVM	0.742	0.689	0.626	0.673	0.658	0.658	0.658
	RF	0.951	0.901	0.882	0.895	0.892	0.888	0.895
	MLP	0.763	0.717	0.680	0.709	0.699	0.700	0.697
	LightGBM	0.937	0.901	0.793	0.858	0.848	0.820	0.885
	AdaBoost	0.687	0.684	0.567	0.652	0.627	0.622	0.632
	XGBoost	0.944	0.906	0.828	0.875	0.867	0.846	0.894
	CatBoost	0.925	0.844	0.833	0.842	0.839	0.840	0.837
Feature selection with t/Chi-Square	LR	0.575	0.462	0.586	0.497	0.523	0.538	0.511
	SVM	0.600	0.519	0.586	0.542	0.552	0.567	0.538
	RF	0.951	0.925	0.818	0.881	0.872	0.841	0.912
	MLP	0.585	0.538	0.591	0.557	0.564	0.579	0.550
	LightGBM	0.871	0.854	0.700	0.797	0.778	0.748	0.821
	AdaBoost	0.685	0.722	0.542	0.668	0.634	0.622	0.651
	XGBoost	0.900	0.906	0.749	0.844	0.829	0.790	0.884
	CatBoost	0.846	0.830	0.724	0.793	0.778	0.759	0.803
Without Feature selection	LR	0.577	0.538	0.571	0.552	0.554	0.567	0.542
	SVM	0.743	0.684	0.680	0.687	0.682	0.690	0.673
	RF	0.951	0.882	0.906	0.895	0.894	0.908	0.880
	MLP	0.797	0.741	0.700	0.730	0.720	0.720	0.721
	LightGBM	0.942	0.892	0.833	0.869	0.863	0.848	0.880
	AdaBoost	0.686	0.660	0.581	0.641	0.622	0.622	0.621
	XGBoost	0.943	0.901	0.818	0.868	0.860	0.838	0.888
	CatBoost	0.936	0.877	0.818	0.855	0.848	0.834	0.865