Figure 6

Accuracy and ROC curves for the KNN classifier. (a,b) Classification accuracy for Country-1 and Country-2 while varying the percentile threshold (i.e., number of features), for R and wSDM (see Supplementary Table 4 for details). (c,d) ROC curves (AUC significance, p < 0.01) [Sensitivity (TPR) vs. 1-Specificity (FPR)] graph for Country-1 and Country-2, for wSDM and R, considering the optimal threshold. The area under the curve (AUC) measures the performance of the classifier across different points of the ROC space. The dashed black line represents random guess (i.e., AUC = 50).