Table 3 Comparisons of different state-of-the-art machine learning models’ performance vs. the proposed model’s performance based on different test data percentages.

Model name	Test data percentage	Precision	Recall	F1 score	Accuracy	Specificity	FPR	FNR	MCC	Marked-ness
Naïve Bayes	10%	0.752	0.743	0.740	0.743	0.63	0.36	0.70	0.49	0.50
	20%	0.759	0.743	0.738	0.743	0.60	0.39	0.12	0.50	0.52
	30%	0.803	0.784	0.779	0.784	0.64	0.35	0.08	0.58	0.61
	40%	0.799	0.780	0.775	0.780	0.64	0.36	0.08	0.57	0.60
Logistic regression	10%	0.752	0.743	0.740	0.743	0.63	0.36	0.15	0.49	0.50
	20%	0.748	0.730	0.724	0.730	0.57	0.42	0.12	0.47	0.50
	30%	0.809	0.784	0.778	0.784	0.62	0.37	0.06	0.59	0.62
	40%	0.808	0.780	0.774	0.780	0.62	0.37	0.06	0.59	0.62
XGBoost	10%	0.715	0.715	0.715	0.710	0.73	0.26	0.30	0.43	0.43
	20%	0.719	0.717	0.719	0.717	0.65	0.34	0.22	0.43	0.43
	30%	0.785	0.775	0.774	0.775	0.69	0.30	0.15	0.55	0.56
	40%	0.768	0.767	0.767	0.767	0.73	0.26	0.21	0.53	0.53
Decision tree	10%	0.672	0.666	0.665	0.666	0.68	0.31	0.30	0.68	0.38
	20%	0.737	0.737	0.737	0.737	0.71	0.28	0.25	0.46	0.46
	30%	0.733	0.733	0.733	0.733	0.73	0.26	0.26	0.46	0.46
	40%	0.751	0.748	0.748	0.748	0.78	0.21	0.28	0.49	0.50
SVM	10%	0.824	0.794	0.789	0.794	0.63	0.36	0.05	0.61	0.65
	20%	0.824	0.794	0.782	0.794	0.63	0.36	0.35	0.61	0.65
	30%	0.845	0.801	0.793	0.801	0.60	0.39	0.01	0.64	0.69
	40%	0.847	0.800	0.791	0.800	0.60	0.40	0.01	0.64	0.70
DNN-A	10%	0.867	0.820	0.810	0.820	0.63	0.36	0.00	0.68	0.74
	20%	0.845	0.807	0.801	0.807	0.63	0.36	0.02	0.64	0.69
	30%	0.896	0.876	0.861	0.871	0.71	0.28	0.00	0.75	0.78
	40%	0.836	0.815	0.816	0.819	0.69	0.30	0.06	0.65	0.67

FPR false positive rate, FNR false negative rate, MCC Matthew correlation coefficient, FPR false positive rate, FNR false negative rate, MCC Matthew correlation coefficient, SVM support vector machine, DNN-A deep neural network-based model with an attention mechanism.

Quick links

Search