Table 3.

Mean 10-fold cross-validation (CV) Training set performance and held-out test set performance of top models from raw and imputed datasets.

Training Performance
	Raw Data		Mean		Median			KNN			Bagged Trees
Top Models	NB	XGBTree	SVM-RB	Log Reg	SVM-RB	Log Reg	DT	SVM-RB	Log Reg	NB	SVM-RB	Log Reg
Brier Score	0.156	0.157	0.117	0.172	0.115	0.251	0.175	0.115	0.228	0.176	0.119	0.190
Accuracy	83.5%	79.9%	86.2%	78.2%	85.4%	74.9%	75.5%	86.3%	72.1%	81.9%	83.5%	78.5%
AUROC curve	0.869	0.861	0.924	0.839	0.919	0.739	0.826	0.908	0.756	0.877	0.909	0.803
Test Performance
	Raw Data		Mean		Median			KNN			Bagged Trees
Top Models	NB	XGBTree	SVM-RB	Log Reg	SVM-RB	Log Reg	DT	SVM-RB	Log Reg	NB	SVM-RB	Log Reg
Brier Score	0.117	0.166	0.069	0.069	0.066	0.066	0.183	0.077	0.077	0.105	0.068	0.068
Accuracy	86.2%	79.3%	93.1%	93.1%	93.1%	93.1%	72.4%	89.7%	89.7%	89.7%	93.1%	93.1%
AUROC curve	0.928	0.856	0.981	0.981	0.986	0.986	0.820	0.976	0.976	0.952	0.962	0.962