Table 5.

Performance metrics of machine learning models with the Synthetic Minority Oversampling Technique.

Algorithms	Accuracy	Area under the receiver operating characteristic curve	Precision	Recall	Brier loss
Logistic regression	98.2	97.4	97.3a	95.9	0.015
Random forest	98.4b	98.0c	96.8	97.3	0.012d
Extra trees	98.1	97.4	97.1	95.8	0.016
Gradient boosting trees	98.1	97.9	95.2	97.7e	0.014
Artificial neural network	96.7	96.2	93.0	95.1	0.026

^aThe highest precision.

^bThe highest accuracy.

^cThe highest area under the receiver operating characteristic curve.

^dThe least Brier loss.

^eThe highest recall.