Adjust shapes at edges of zipcode domain to match zips instead of all pumas

[famulare]

Compare zipcode boundary https://github.com/seattleflu/seattle-geojson/blob/master/sfs_domain_geojsons/sfs_domain_zipcodes.geojson

to puma-matched boundary https://github.com/seattleflu/seattle-geojson/blob/master/sfs_domain_geojsons/sfs_domain_neighborhood%2Bpuma.geojson

The puma matching puts a lot of visual mass on low density populations we aren't really serving. We should consider restricting the boundaries to the zip code edge.

The potential downside is making it harder to match to census data, but if we do that at the tract level, I don't think it's a problem.

[tsibley]

Does adjusting these shapes affect the model computation, or is the goal solely to make display better?

[famulare]

Display only.

[famulare]

For now anyway. If/when doing more with census, we just need to define things clearly.

[famulare]

@tsibley I uploaded shapes with better rural boundaries. Copying slack comment here too: https://app.slack.com/client/TCAKABJR2/CE9B78CS0/thread/CE9B78CS0-1575587566.079000

I think they'll be good enough to display, but I defer to you on UI/UX if you still aren't happy with the viz.

https://github.com/seattleflu/seattle-geojson/blob/master/sfs_domain_geojsons/sfs_domain_neighborhood%2Bpuma.geojson

We can further refine these if we want to, but no time before launch. I did the clipping based on zipcodes, and that left a few edge cases that I don't like (such as including Tiger Mountain where no one lives but excluding Snoqualmie and North Bend). But these cover 95%+ of the SFS samples from year 1

[tsibley]

Thanks! These definitely look better. I've switched the map over to them. :-)

I would still like to clip these (at least for display) to not go over water as well.

[famulare]

Will remove water before Monday morning.

[famulare]

Done. Cropped out Lakes Union, Washington, and Sammamish.