Geocoding 101

a crash course in turning localities into lat/lngs

 

 

What

is

Geocoding?

Street_Address City State Zip_Code
3250 LAKESHORE AVE OAKLAND CA 946102720
17035 LAUREL RD MORGAN HILL CA 950374415
635 COLEMAN AVE SAN JOSE CA 951102029
2100 SANTA ROSA AVE SANTA ROSA CA 954077691
3649 MT DIABLO BLVD LAFAYETTE CA 945493735
5000 FOLSOM BLVD SACRAMENTO CA 958194441
9670 BRUCEVILLE RD ELK GROVE CA 957575950

Why Geocode?

What else is possible with this type of data?

Route Optimization

Origin-Destination Matrices

Service Areas & Travel Time Isohyets

Cost
Scale

Rights

How

is

Geocoding

Done?

Census Use Study of 1967

(we like a bit of history, right?)

Network-based

(% along geocoding)

25 Main Street, Dallas, TX 75208

how do you put that on a map?

Search-based geocoding 

Geocoding Infrastructure

The New locator.stanford.edu Service

Bulk Geocoding

Gazetteering

Reverse Geocoding

1.67 million geocodes per hour

Geocoding APIs & Services

Specialty Geocoding APIs:

  • Yandex - Geocoding and Routing API for Russia, Eastern Europe and Western Asia.
  • MapMyIndia.com - Geocoding and Routing API (plus many others!) for India

Esoteric localities...

Chambers Spring Road 2.5 km S of Hwy 412, 8.0 km E of Siloam Springs, T17N, R33W, Sec. 18

Tulane's Geolocate Infrastructure can handle this...

http://www.geo-locate.org/

Most important, Geolocate emphasizes the idea of 'uncertainty' in geocoding

http://www.geo-locate.org/web/WebGeoref.aspx

The Data

250+GB of Parcel, Street, POI & Administrative Boundary Data

Street-level address data for 149 countries that covers more than 90% of the world's population

Millions of businesses, landmarks, and other points of interest for "Reverse Geocoding"

Topologically Integrated

Thinking about geocoding...

Things GIS hates about your tables:

  • Spaces or special characters (other than _ ) in field names
  • Field Names that start with a number
  • Empty Rows
  • Merged Cells
  • Excel Formulas

Specifics on tabular data for geocoding

  • Most geocoding services prefer street addresses in a single field, with City, State, Postal Codes, etc... in separate fields.
  • For placenames, concatenate administrative units, as follows:

Dallas, Tx, USA

Collapse your addresses/placenames into a table of UNIQUE addresses/placenames!

https://locator.stanford.edu/arcgis/rest/services/Geocode/NorthAmerica_Composite/GeocodeServer/findAddressCandidates?f=pjson&singleLine=397%20Panama%20Mall,%20Stanford,%20CA%2094305

https://locator.stanford.edu/arcgis/rest/services/Geocode/NorthAmerica/GeocodeServer/findAddressCandidates?f=pjson&singleLine=397%20Panama%20Mall,%20Stanford,%20CA%2094305