TABLE 3.

Dietary biomarkers ensemble classifiers’ results. Highest precision, recall, F2-score and AUC reached for each algorithm: Decision Tree (DT), Logistic Regression (LR), Naïve Bayes (NB), Neural Network (NN), Random Forest (RF) and Support Vector Machine (SVM). The NB algorithm did not work with the Stacking approach.

BAGGING
	MaxPrecision	MaxRecall	MaxF1	MaxF2	Max-AUC
DT	0.742	0.477	0.546	0.502	0.941
LR	0.664	0.707	0.642	0.663	0.950
NB	0.716	0.580	0.594	0.582	0.951
NN	0.729	0.477	0.537	0.497	0.942
RF	0.648	0.541	0.571	0.550	0.958
SVM	0.740	0.405	0.489	0.433	0.957

STACKING
	MaxPrecision	MaxRecall	MaxF1	MaxF2	Max-AUC
DT	0.417	0.735	0.513	0.622	0.842
LR	0.380	0.890	0.521	0.685	0.961
NN	0.581	0.505	0.521	0.509	0.911
RF	0.568	0.734	0.624	0.673	0.952
SVM	0.374	0.890	0.513	0.679	0.947

The highest value of each metric is bolded.