Fig. 3
From: Determining molecular properties with differential mobility spectrometry and machine learning
Random forest regression ML fits for CCS and pKb along with learning curves. Learning curves for CCS a and pKb c show the evolution of mean absolute error for test set (black) and training set (red) as the size of the ML training set is increased. MAE for test set is shown in black while that for training set is shown in red. Random forest results for CCS and pKb are shown in b and d, respectively. Individual points are colored based upon the molecular set as indicated in the figure. The red solid line in b and d is a plot of y = x. Error bars (2σ) are calculated from the standard deviation of the fitted parameter
