Table 7.

Predictive performance of dynamic behavioral models for violent behavior.

Model type	Sensitivity	Specificity	Positive predictive value	Negative predictive value	Accuracy	AUC
MLP	0.6216	0.7612	0.5897	0.7846	0.7115	0.7370
RF	0.6486	0.7612	0.6000	0.7969	0.7212	0.7475
KNN	0.2703	0.9851	0.9091	0.7097	0.7308	0.6846
SVM	0.5405	0.8209	0.6250	0.7639	0.7212	0.7348
LR	0.5135	0.8507	0.6552	0.7600	0.7308	0.8003
XGBoost	0.4865	0.8060	0.6061	0.7606	0.7115	0.7402