Closed elray1 closed 3 months ago
I see your point and agree that some rewording is necessary. One concern I have, though, about not mentioning it is that the resulting forecasts queried from zoltar (which are stored in data objects in the hubEnsemblesManuscript repo) have some column names for the task IDs that differ from the forecasts in the FluSight repo, which may be confusing to anyone accessing the raw forecasts associated with the manuscript. However, I'm fine with changing the second sentence to something more general like "Forecasts must conform to hubverse standards to be fed into either of the ensembling functions, so we first transform the raw forecasts using as_model_out_tbl
..."
I also wanted to note that queried the forecasts from zoltar because of issues with the load_forecasts
function in covidHubUtils, which is shown in some of the scripts provided in the manuscript repo, and I didn't want to misrepresent my process. One way around defining stuff about zoltar in the manuscript could be adding comments in the script that contains this code itself. Any thoughts or preferences about this solution?
Thanks for providing more information about this. I'm ok with just leaving the references to Zoltar in if that's an easier and more accurate route forward.
your more general sentence does seem good, though.