Table 3.

SVM and k-NN classifier’s performance on the test set.

Classifier	ROC AUC A	F1 Score A	Balanced Accuracy A	MCC A
MAP4 SVM B	0.97	0.91	0.93	0.86
MAP4 k-NN C	0.96	0.88	0.90	0.81
Physchem SVM D	0.86	0.73	0.78	0.56

^A Area under the receiver operating characteristic curve (ROC AUC), F1 score, balanced accuracy, and MCC are metrices used to evaluate a machine learning model. MCC can assume values from –1 to 1, all other parameters can assume values from 0 to 1, and in all cases 1 is a perfect classification. Refer to Section 2 for details. ^B SVM classifier trained with the MAP4 fingerprint. ^C k-NN classifier trained with the MAP4 fingerprint. ^D SVM trained with physiochemical properties.