Figure 3.

Learning curves with $F_1$ score for multi-model/multi-label methods trained on imbalanced dataset. The dotted lines show the total max/min score across all CV-repeats for each training set size. The thick solid line shows the median score. The shaded area marks the median average deviation (across all CV-repeats) and contains $\ge \mathrm{50\%}$ of scores. For exact numbers and confidence intervals see Suppl. Tables S3 and S4.