search

epiforecasts / RtD3

Rt visualisation in D3
https://epiforecasts.io/RtD3/
Other
6 stars 2 forks source link

Link via region codes and not names. #6

Closed seabbs closed 3 years ago

seabbs commented 4 years ago

It might make sense to link via region codes and not names as this appears to be causing some issues in practice. covidregionaldata should have region codes for all regional data which might make this easier for our estimates. I imagine it might also be easier for other users.

hamishgibbs commented 3 years ago

I believe that with the variety of region codes that exist, in the UK for example, local authorities, NHS regions, counties, GADM data, or rnaturalearth data, which all have different reference codes - it may be easiest to keep relying on matching names (and warn when some names will be missing from the output).

At the moment, the names in the dropdown are defined by the names in the time series data, not the map data, so if there are disagreements, they will still be accessible from the dropdown but will be missing from the map. This seems like a good approach to ensure that the rt data remains accessible.

Admin name wrangling is painful but I believe that finding geodata that aligns with different region codes may be more painful, especially because of the ease of using rnaturalearth.

So I support continuing to rely on admin names at the moment, but would like to hear your perspective.

hamishgibbs commented 3 years ago

@seabbs are you ok to close this? and continue with names for the time being?

seabbs commented 3 years ago

Yes I think so. My only concern is that it was a little painful linking some countries up (i.e here: https://github.com/epiforecasts/covid/blob/master/_posts/national/colombia/colombia.Rmd). However, perhaps there is no way to resolve that....