Table 3.

Model performance for the 4 model types (DT, RF, ANN, and SVM), with the 2 sampling versions (undersampled, oversampled).

Model	Version	Train accuracy	Test accuracy	Test precision	Test recall	Test F1	Test AUC
DT	Undersampled	1.0	0.74	0.68	0.75	0.71	0.837
DT	Oversampled	1.0	0.73	0.66	0.71	0.69	0.760
RF	Undersampled	1.0	0.80	0.78	0.74	0.76	0.868
RF	Optimized	1.0	0.81	0.80	0.74	0.76	0.863
RF	Oversampled	1.0	0.81	0.82	0.70	0.75	0.866
ANN	Undersampled	0.77	0.76	0.71	0.73	0.72	0.830
ANN	Oversampled	0.66	0.61	0.52	0.96	0.68	0.842
SVM	Undersampled	0.70	0.68	0.59	0.84	0.69	0.724
SVM	Oversampled	0.71	0.69	0.60	0.83	0.69	0.754

Final optimized RF model outcomes are provided.

Abbreviations: ANN: artificial neural networks; DT: decision trees; RF: random forest; SVM: support vector machines.