No ili data from FL

[stephenturner]

Oh boy. Running into issues at #7 (#23). It looked to be pretty straightforward to add location as a key in the tsibble, then everything "just works" downstream. Except that all states are missing weighted ILI data. I thouight we'd move to unweighted ILI, but Florida is missing all unweighted ili and ilitotal (so is the Commonwealth of the Northern Mariana Islands, but it's being tossed in a join against location data that doesn't have this territory by name).

library(fiphde)
x <- get_cdc_ili(region=c("national", "state"))
x %>% 
  group_by(region) %>% 
  summarize(missing_wili=sum(is.na(weighted_ili)), 
            missing_uili=sum(is.na(unweighted_ili)), 
            missing_ilitotal=sum(is.na(ilitotal))) %>% 
  arrange(desc(missing_uili), desc(missing_wili)) %>%
  head()

# A tibble: 6 × 4
  region     missing_wili missing_uili missing_ilitotal
  <chr>             <int>        <int>            <int>
1 Florida             584          574              574
2 Alabama             584            0                0
3 Alaska              584            0                0
4 Arizona             584            0                0
5 Arkansas            584            0                0
6 California          584            0                0

You can tear this back down to cdcfluview::ilinet() and you get the same thing.

Here's my suggested path forward:

1. Switch to unweighted ILI. We might not have a FL forecast on Jan 10, and that might be okay.
2. Alternatively, and I don't know if this would actually work, we could get data from cdcfluview::ilinet at the HHS regional level, looking only at HHS region 4, then subtract out Georgia, AL, NC, SC, etc to get at Florida? This seems like a path laden with gotchas.
3. Find some alternative data source just for florida.

[stephenturner]

This was mostly dealt with in #23 - I get all states' unweighted ILI, except florida, which has no data. This is more of a modeling thing that'll have to happen in #33, in that if we're going to produce state-level forecasts and we have no data on florida, we'll have to deal with it there, not necessarily here. Closing this one for now.