search

nvkelso / election-geodata

Precinct shapes (and vote results) for US elections past, present, and future
319 stars 51 forks source link

Add 2016 election precinct shapes for Kansas #9

Closed nvkelso closed 7 years ago

nvkelso commented 7 years ago

This state is in priority tier 1 (out of all US states) according to https://lawreview.uchicago.edu/sites/lawreview.uchicago.edu/files/uploads/82_2/04%20Stephanopoulos_McGhee_ART.pdf (via http://mike.teczno.com/notes/redistricting/measuring-efficiency-gap.html).

105 counties. _Worse than California, only better than Texas, Georgia, Virginia, Kentucky, and Missouri. src_

nvkelso commented 7 years ago

When researching & requesting data also ask for matching election results (presidential, congressional, state house, and state senate) in addition to the precinct shape data for 2016, and for earlier election years if available.

nvkelso commented 7 years ago

Looks like incomplete county based precinct coverage at Open Elections:

karpet commented 7 years ago

Great project. I am investigating KS precinct GeoJSON data, which appears to be necessary county-by-county. Starting now with Douglas County.

nvkelso commented 7 years ago

@karpet Thank you! I've updated the issue description to include checkboxes for the 105 counties of Kansas and their FIPS codes (src).

It looks like US Census has a list of voting precincts (last decennial in 2010) in all of Kansas to give us a rough idea of total coverage count. There are 3907 in that file (linked below).

The Kansas precincts mostly have sane names (versus California where most are just numbered), I wonder if they are reverse geocode-able? Many look to correspond to localadmin areas, of which Who's On First (via US Census) has 1530 count and polygons (which are downloadable).

When I do some quick text analysis it look like a pretty good match, with the other ~1500 precincts being in urbanized city areas where we could at least get a rough "city" lat/lng point, and a more specific polygon (or centroid) for the townships. This iterative approach might be useful to prioritize which counties have more-or-less complete township === precinct data, versus mostly urban counties where the more detailed precincts would be useful. Population count is also good priority proxy.

@migurski do you need polygons or would lat/lng points work in a pinch? Over time more would be polygons, but looks like right now we could do 50% of one, 50% of the other.

STATE|STATEFP|COUNTYFP|COUNTYNAME|VTDST|VTDNAME
KS|20|145|Pawnee County|000050|Grant Township Voting District
KS|20|145|Pawnee County|000060|Keysville Township Voting District
KS|20|145|Pawnee County|000070|Larned Township Voting District
KS|20|145|Pawnee County|000080|Larned Ward 1
KS|20|145|Pawnee County|000090|Larned Ward 2
KS|20|145|Pawnee County|000100|Larned Ward 3
KS|20|145|Pawnee County|000110|Larned Ward 4
KS|20|145|Pawnee County|000120|Lincoln Township Voting District
KS|20|145|Pawnee County|000130|Logan Township Voting District
KS|20|145|Pawnee County|000140|Morton Township Voting District
migurski commented 7 years ago

I think points could work — we could use them to build a Voronoi tesselation as a cheap starter set of polygons, and then refine if more data becomes available? Really curious if @anthea is still present on Github and has something to say about this.

nvkelso commented 7 years ago

Using Pivot table in OpenOffice I started with the text list from Census, created a new calculated field that's FIPS plus county name + State 2 char code on the left, and created a new "count" column I added to the rows with every row with 1 on the right. I then sum up the precinct counts per county using the pivot table. (Pivot tables in OpenOffice show up below the current selection, so scroll way down.)

I've updated the issue description to include precinct counts per county (but kept that list alphabetical).

Here are the counties with 50 or more precincts and where I'd prioritize work:

FIPS County Precincts
20173 Sedgwick County, KS 447
20091 Johnson County, KS 444
20177 Shawnee County, KS 227
20209 Wyandotte County, KS 111
20045 Douglas County, KS 95
20015 Butler County, KS 81
20155 Reno County, KS 80
20161 Riley County, KS 71
20035 Cowley County, KS 69
20169 Saline County, KS 64
20037 Crawford County, KS 62
20061 Geary County, KS 61
20125 Montgomery County, KS 60
20121 Miami County, KS 59
20021 Cherokee County, KS 54
20111 Lyon County, KS 53
20103 Leavenworth County, KS 52
20009 Barton County, KS 50
karpet commented 7 years ago

@nvkelso I got the ERSI geodata file from the county, converted it to PostGIS, and then exported it as GeoJSON. Not sure if it's the format you're looking for though. Happy to transform it into something compatible if you can point me at a schema or example.

https://github.com/karpet/election-geodata/tree/master/data/20-kansas/20045-douglas

nvkelso commented 7 years ago

For those like me catching on slowly, Douglas Co = Lawrence. Kansas :)

I loaded your GeoJSON into QGIS and it shows up in the right spot, with good attributes. Works for me!

For some reason Github won't preview the GeoJSON file, but the validation in QGIS shows it's good enough.

Want to file a PR for this one from your fork?

On Tue, Mar 7, 2017 at 1:03 PM, Peter Karman notifications@github.com wrote:

@nvkelso https://github.com/nvkelso I got the ERSI geodata file from the county, converted it to PostGIS, and then exported it as GeoJSON. Not sure if it's the format you're looking for though. Happy to transform it into something compatible if you can point me at a schema or example.

https://github.com/karpet/election-geodata/tree/master/ data/20-kansas/20045-douglas

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/nvkelso/election-geodata/issues/9#issuecomment-284858171, or mute the thread https://github.com/notifications/unsubscribe-auth/AA0EO6h8qCRytel3hN_EXqKlCpoAX0aDks5rjcYhgaJpZM4MTtm1 .

karpet commented 7 years ago

Yup #55

nvkelso commented 7 years ago

Progress!

image

karpet commented 7 years ago

I was able to locate statewide districts for 2010. Anything newer will require going county-by-county, but I am discovering the GIS community in KS is happy to help when asked. Ever forward.

See #60

migurski commented 7 years ago

So glad to hear it, @karpet!

nvkelso commented 7 years ago

Looks like 2012 precincts may also be available:

karpet commented 7 years ago

Confirmed.

http://kansasgis.org/catalog/index.cfm?data_id=2081&show_cat=1?data_id=2081&show_cat=1

I will add to #60

nvkelso commented 7 years ago

Closed #9 via https://github.com/nvkelso/election-geodata/pull/60.