Fig. 3

An overall comparison between the performance of the five machine learning methods used for the construction of the off-target models, an overview of all the performance metrics and the balanced accuracy ranges for the neural networks. a A bar plot comparing Neural Networks, H2O, AutoGluon, RandomForest and Auto-Sklearn (x-axis) with respect to the number of targets each method scored the highest Mathews Correlation Coefficient(MCC), Balanced Accuracy(BA) and F1 (y-axis).The bars are color coded according to the machine learning method and the number of the targets each method scored highest is indicated on the top of each bar. b A box plot comparing the values (y-axis) of the different performance metrics (x-axis) for the Neural networks method. c A bar plot displaying the number of off-target models (y-axis) falling under each balanced accuracy range (x-axis) for the neural networks method