Braindr rating aggregation and rater importance: A. ROC curves on the test set for various training set sizes (here n denotes the number of slices used). The dashed line is the ROC curve of the average crowd rating for all slices. B. The distribution of XGBoost probability scores on all Braindr slices. C. Feature importance for each anonymized user. D. Relationship between rater importance and total number of ratings in the gold standard dataset.