Fig. 3
From: Screening PubMed abstracts: is class imbalance always a challenge to machine learning?
Forest plots of delta AUCs by balancing and machine learning techniques (MLTs). Forest plots that show differences in AUC (delta AUCs) between the AUCs obtained with each balancing technique (i.e., RUS-50:50, RUS-35:65, ROS-50:50, and ROS-35:65) and the AUC obtained without the application of any of them for each combination of MLT and systematic reviews. Red diamonds report to pooled results obtained with a by-MLT meta-analytic fixed-effect model. The first author and year of systematic review corresponding to each row of the forest plots are reported in the first column only, the MLTs are reported in the first row only, and the balancing techniques are reported in each forest plot’s x-axis label