Table 2.

Results of the model selection procedure on the INN development set [N = 23758].

MODELS	TRAINING	CROSS-VALIDATION
Logistic Regression	0.9105 ± 0.0010	0.9098 ± 0.0038
K-Nearest Neighbor	0.9142 ± 0.0012	0.9108 ± 0.0040
Random Forest	0.9373 ± 0.0010	0.9138 ± 0.0053
Gradient Boosting Machine	0.9200 ± 0.0011	0.9147 ± 0.0048
Support Vector Machine	0.9170 ± 0.0013	0.9147 ± 0.0047
Neural Network	0.9171 ± 0.0010	0.9149 ± 0.0047

For each candidate model, its training and validation AUROC (averaged over 5 CV iterations, ±standard deviation) is reported. The selected model is highlighted in bold.