Get flusight bin endpoints

[elray1]

This contains a general setup that I think should work for both seasons, but I ran out of time for testing the stuff for the 22/23 season, so I mostly took it out and dropped code into #16. There are a couple of vestigial leftover bits that I didn't want to delete since we will want to recreate it at some point...

[lshandross]

I'm approving because everything looks good to me, but I have one minor question: why did you decide to make the sign for the bin endpoints separate columns?

[elray1]

Thanks!

why did you decide to make the sign for the bin endpoints separate columns?

I basically felt that this was clearer when working with the pmax operation here. If the sign is negative, what you would want is to add pmin(.data[["lower_rate_multiplier"]] * .data[["pop100k"]], .data[["lower_min_count_change"]]). But you'd also need to make sure that the sign was the same for both of those arguments. And... would it be the same pmin/pmax decision for both the lower and upper endpoints? The answer is yes, but I coded it up wrong in my first try.

[lshandross]

One other thing to keep in mind — FluSight had slightly different category definitions for the first few weeks of the 2023/2024 season, which were then adjusted at the beginning of November 2023. I don't know if it makes sense to have an option to use these in the function to get been endpoints for the 2023/2024 season, but it's also possible this could account for some of the mismatches between quantile forecasts and the categorical ones.

[elray1]

I didn't remember that. do you know of a place where those different category definitions are recorded?

[lshandross]

Here's a link to the specific section in the model-output README before the categories were updated in the FluSight-forecast-hub repo

[elray1]

Thanks, just noting that I filed issue #17 to address this point!