Evaluating machine learning-powered classification algorithms which utilize variants in the GCKR gene to predict metabolic syndrome: Tehran Cardio-metabolic Genetics Study

Table 4 Performances metrics for LR, SVM, DT, RF, LDA, and QDA algorithms

	Models	Accuracy	Sensitivity	Specificity	Kappa	AUC-ROC	AUC-PR
Total	SVM	0.725	0.661	0.785	0.447	0.785	0.761
	DT	0.738	0.667	0.804	0.473	0.771	0.730
	RF	0.743	0.699	0.784	0.484	0.804	0.776
	LR	0.705	0.677	0.732	0.409	0.770	0.748
	LDA	0.562	0.915	0.230	0.141	0.658	0.666
	QDA	0.546	0.492	0.598	0.089	0.563	0.555
Male	SVM	0.712	0.475	0.870	0.366	0.733	0.766
	DT	0.735	0.527	0.874	0.421	0.739	0.753
	RF	0.729	0.559	0.842	0.415	0.754	0.782
	LR	0.711	0.519	0.839	0.373	0.732	0.768
	LDA	0.591	0.754	0.482	0.217	0.679	0.734
	QDA	0.547	0.394	0.649	0.044	0.531	0.616
Female	SVM	0.733	0.783	0.671	0.456	0.802	0.565
	DT	0.748	0.753	0.742	0.492	0.785	0.706
	RF	0.744	0.767	0.715	0.482	0.815	0.754
	LR	0.738	0.798	0.663	0.465	0.803	0.741
	LDA	0.608	0.752	0.427	0.184	0.664	0.617
	QDA	0.635	0.749	0.491	0.245	0.670	0.572

LR Logistic Regression, SVM support vector machines, DT Decision Tree, RF Random Forest, LDA Linear discriminant analysis, QDA Quadratic discriminant analysis, AUC Area Under Curve. Machine learning methods outperforms the traditional statistical methods

ISSN: 1479-5876