Table 5.

Prediction accuracy, F1 score, and AUC for all algorithms trained and tested on the Canadian dataset. Accuracy was evaluated by the average of a three-fold cross validation

Model	Accuracy	F1	AUC	3-,Accuracy	3-,F1	3-,AUC
Logistic regression	0.919	0.872	0.917	0.957	0.965	0.951
Random forest classifier	0.904	0.846	0.924	0.963	0.968	0.956
K-Neighbors classifier	0.830	0.712	0.881	0.909	0.928	0.897
SVC	0.926	0.885	0.918	0.959	0.967	0.953
Gaussian process classifier	0.919	0.872	0.916	0.954	0.962	0.947
AdaBoost classifier	0.889	0.831	0.926	0.957	0.965	0.956
XGB classifier	0.933	0.895	0.926	0.946	0.956	0.939

Note: We compare the entire dataset (538 cases) with learning and testing on data with no more than

three missing variables (labeled 3-, 466 cases)