ROC curve for SNGuess with the test data set (see Sect. 2.2). The area under the curve (ROC AUC) value of 0.93 summarizes a good performance across different score thresholds for distinguishing between relevant and nonrelevant candidates. As a comparison point, we also see the ROC curve for performing a simple logistic regression classification with just the distance to nearest source as an independent variable. We can see that SNGuess shows a better performance than this simple classification (ROC AUC = 0.66).