Fig. 4From: Old drug repositioning and new drug discovery through similarity learning from drug-target joint feature spacesHistograms of precision/recall/F1-Score/AUC values for four classical classifiers on reliable, pairwise, and randomly generated negative samples. Panel (a) shows the precision, panel (b) shows the recall, panel (c) denotes the F1-Score and panel (d) is the AUCBack to article page