Per @elray1 's suggestion, I gave the manuscript a fairly high level overview since a few others have already gone through and given detailed comments. Overall, I think the manuscript looks great. No major comments or suggestions at this point. I do have a few more minor comments for consideration in a future draft.
I wonder if it might make sense to spend just a bit more space describing the hubverse concepts. a small table that shows the output_types and describes what they are? A tabular (not R output) version of some hubverse data that shows the canonical columns and then the task-id columns? It would help clarify these concepts, and since we haven't really documented those concepts "in the literature" anywhere, this might be as good a first place for them as any.
For the examples that show data as raw R output, I suggest filtering to a smaller number of quantiles, so a small number of rows (I think only six are shown via head() currently) can show an entire "prediction" and not just through quantile 0.2. Maybe 7 rows with the median and three intervals that are shown in the plots? I think this would make the example "whole" for people who are less familiar with the data structure and would allow a reader to follow the data for an entire single prediction through the tables and into the figure.
Per @elray1 's suggestion, I gave the manuscript a fairly high level overview since a few others have already gone through and given detailed comments. Overall, I think the manuscript looks great. No major comments or suggestions at this point. I do have a few more minor comments for consideration in a future draft.
head()
currently) can show an entire "prediction" and not just through quantile 0.2. Maybe 7 rows with the median and three intervals that are shown in the plots? I think this would make the example "whole" for people who are less familiar with the data structure and would allow a reader to follow the data for an entire single prediction through the tables and into the figure.