Description: Zip_Backfill.py takes an OpenAddresses files and, for rows that do not have zip codes, compares the row's lat and lon to a shape file of zip codes in order to assign it a zip code. The code is designed to work with a Census ZCTA .shp file.

All other files are to test the efficacy of Zip_Backfill.py.

• To install dependencies run pip install -r requirements.txt

• Download the 2015 TIGER ZCTA zip here

• Retrieve the OpenAddresses US regional zip files

• To Backfill

  • Pick an OpenAddresses file from one of their regional US zip files to backfill with zip codes

  • Run python Backfill_Zip.py <CLI Options>

• To Run Tests

  • Make a file of unedited OpenAddresses rows to test with either all or no zip codes

    • To do this run python pull_test_rows.py <CLI Options> to pull rows without zip codes(for a check against Google or Mapbox)

    • OR run python pull_test_rows.py <CLI Options> -z to pull rows with zip codes(for a check against zip codes already in OpenAddresses)

  • Run python test.py <CLI Options>

• Backfill_Zip.py

  • input OpenAddresses file to backfill from
  • output file to to backfill to
  • shape_file a census ZCTA shape file

• pull_test_rows

  • input A file containing some number of state OpenAddresses files
  • output The file to store found rows
  • -z, --zips Pull only rows WITH zip codes instead of rows without zip codes

• test.py

  • input OpenAddresses file to take rows from
  • shape_file a census ZCTA shape file
  • number the number of rows to test
  • -g, --google use google geocode search, argument: (output file)
  • -r, --reverse use a reverse google geocode search, argument: (output file)
  • -z, --zip search google with zip code, argument: (output file)
  • -m, --mapbox reverse geocode from the mapbox API, argument: (output file)
  • -o, --openaddresses check against zips in openaddresses, argument: (output file)
  • -gk, --google_key a file with a google API key, needed if running any check against google
  • -mk, --mapbox_key a file with a mapbox API key, needed if running a check against mapbox

Rows randomly pulled from all files and using the Census Tiger ZCTA .shp file:

• 19,729/23,000 = 86%: backfilled

• 1,608/1,750 = 92%: agree with Google api

• 8,587/8,593 = 99% agree with Mapbox api

• 8,236/8,581 = 96% agree with OpenAddresses