Table 4.

Metrics and performance comparison among ML algorithms for predicting the presence of Pre-MetS and MetS

Model	status	Accuracy	Specificity	Sensitivity	Precision	F1 scores	Brier loss
	Pre-MetS
LR		0.705	0.901	0.395	0.730	0.513	0.202
KNN		0.639	0.797	0.395	0.558	0.463	0.212
GNB		0.729	0.770	0.667	0.653	0.659	0.192
SVC		0.655	0.837	0.375	0.600	0.461	0.232
GBC		0.656	0.840	0.396	0.594	0.475	0.207
VE		0.730	0.905	0.438	0.758	0.571	0.191
	MetS
LR		0.730	0.813	0.629	0.736	0.678	0.170
KNN		0.744	0.720	0.774	0.696	0.732	0.150
GNB		0.778	0.827	0.709	0.771	0.739	0.195
SVC		0.766	0.787	0.741	0.741	0.732	0.164
GBC		0.773	0.787	0.661	0.803	0.726	0.148
VE		0.781	0.853	0.693	0.797	0.741	0.142

LR Logistic Regression for Classification, GNB: Gaussian Naive Baye: KNN: k-nearest neighbour classification, SVM: GBC: Gradient Boosting Classification, Support Vector Machine, VE: Voting Ensemble