Figure 1.

Learning curves with $F_1$ score for models trained with different algorithms on balanced subsets of the dataset. The dotted lines show the total max/min score for each training set size across all subsets and CV-repeats. The solid lines (one per subset) represent elementwise median of curves for all CV-repeats. The thick line is the elementwise median of the 11 median curves shown. The shaded inner area contains all curves plus/minus their median average deviation (across all CV-repeats), and marks a range of the typical performance. For exact numbers and confidence intervals see Suppl. Tables S1 and S2.