Fig. 4

Plot of the difference of classification model accuracies between MA- and individual-classification approach in the simulated datasets, when Δ = 0.1, γ = 0.75 and (a) n = 50 (Simulation 1) (b) n = 100 (Simulation 4) (c) n = 150 (Simulation 7). The aforementioned simulation parameters resulted in the less informative datasets