Geocoding

GIS Workshop
Geocoding – Revisited
Chieko Maene
c‐maene@northwestern.edu
Maps and State Documents Librarian
Government  and Geographic Information and Data Services
University Library (Evanston)
Northwestern University
• Geocoding:
– The process of assigning a pair of approximate
geographic coordinates to each case (with
spatial information, such as: address, zipcode or
place name,
name etc.)
etc )
– Two elements: Source & Reference files
• Address matching
– Most common geocoding method – matching
street addresses with reference street data
– Address matching methods vary by country

How Does Geocoding Work? (US Streets) Geocoding Problems and Solutions
Original Address
• New or irregular street addresses Good reference data and
1745 Sherman Ave, Evanston, IL, 60201 algorism can handle known
– New subdivisions and rural roads
TO 1799 problems, such as ArcGIS and
TO 1798 – Non-grid street system (Washington DC) ESRI StreetMap data
Parsed Address
– International countries (Ex. Japan – street addresses are assigned
1745 | Sherman | Ave | Evanston | IL | 60201
chronologically)
Left side Right side
street street • Expensive street reference data
Street database (reference file) looks for best numbers numbers
matching record(s) – International street (very expensive to obtain..)
 Find streets (& type): Sherman (Ave) 1745 Sherman Ave Solutions
 Find street segments: 1745 falls between 1701 –
1799 (right side of the street line segment) • Instead of street, match by zipcode, or city centroid (see next)
 Calculate/interpolate a geographic coordinate of – Lower accuracy
the street number, 1745, base on the range of the
street line segment (plus a given offset distance) FROM 1700 FROM 1701
• Parcel matching
– Parcel data required (be aware: house numbering mismatch cases)
• Geocode with GPS (latitude & longitude information)
– Locations were recorded with GPS – highest accuuracy

Geocoding Process in ArcGIS Geocoding Variations

Reverse Geocoding
1. Build a geocoding service (a.k.a. geocoder, locator) Reverse Geocoding (New!) Pick the closest street address by
clicking on a map
– Define a method, reference data & fields, (quality)
thresholds/criteria, and output data 1. “I cannot geocode a location because
there is no streets nearby, but I know
2. Run the service against your table data exactly where it is.”
3 Manual intervention/corrections
3. – We can assign geographic coordinates (or
sometimes the closest street address) by
– Manually match unmatched cases that didn't satisfy the defined clicking on a map .
service thresholds/criteria
2. "I know geographic coordinates but I want
4. Try a lower-quality geocoding service to achieve a to know a street address for this location
higher geocoding rate (or a name of the city or zipcode.)"
– [Optional: Use this option with caution] – By submitting geographic coordinates, some
system can assign the closest street
– Geocode based on second best spatial information address, city and/or zipcode.
• i.e. zipcode, instead of street addresses

1
 Exercise 9 Exercise 9 (cont.)
• Let’s geocode campus area restaurants - • Adding X/Y data
• Sometimes your data comes with handy X/Y (Longitude/Latitude
geographic coordinate) information, such as GPS

Reference Data Sources Reference Data Sources (cont.)

• The traditional geocoding services require: • Source options:
• Reference (geocodable streets) data – TeleAtlas
• Source options: • Geocode ready
• Available from ESRI Data & Maps (free for us/NU)
– Census TIGER/Lines – Based on 2003 data
• Free and readily available • Also available from ESRI StreetMap Premium
– 2000 – Better/improved data
– 2008 and later (improved quality) – Updated often
• Data must be processed • Used by Google Maps
– The 2008 data needs to be processed to be used as a
geocoding reference data source – NavTec
• TeleAtlas's rival company
• Used by Yahoo Maps, otherwise expensive to access

Geocoding Variations Exercise 10

• Zipcode/City matching Zipcode data matched with Which is correct?
centroid of zipcode polygons • Geocoding of U.S. Embassies • Florence, Italy
• “My data doesn’t have street level • Firenze, Italy
address but has a zipcode field” – Note: geocoding international
• We can match the data with zipcode places is not easy because of
points or polygons non-standarized naming and
• Two
T elements
l t foreign characters (unicode)
– Source file (i.e. data with zipcode)
– Reference file (zipcode point or polygon
data: either way, output will be points)
• Tools: ArcToolbox > Geocoding Tools
1. Create Address Locator
2. Geocode Addresses

For more information:

http://webhelp.esri.com/arcgisdesktop/9.3/index.cfm?TopicName=Commonly_used_address_locator_styles

2
 New: Online Services in ArcGIS ESRI Online Locators
http://www.esri.com/software/arcgis/arcgisonline/world-geocoding.html
• Use of pre-defined online geocoding services in
ArcGIS • Features:
• Pros – Good! – Free (or subscription for better services)
– No reference (street) data is necessary – Data updated frequently

• Cons – Bad! • Type:

T
– Not much control over the results – careful follow-up (i.e. – World Gazetteer
quality check) would be necessary – US and North American
• Most services typically use "composite" geocoding service
(i.e. A "composite" service tries to use the "best" spatial • Use:
information but if the first choice doesn't work, they will use – Find
the second best information, such as geocode by zipcode, by
place and then by state.) – Batch geocoding (up to 1,000 per session)
To add ESRI Online Locators, find and select GIS Servers >
Add ArcGIS Server > Use GIS Services, and then add
http://tasks.arcgisonline.com/arcgis/services in server URL

Demonstration Reference
Geocoding using Online Service

• In the real world, you need to • For more information: Geocoding in ArcGIS 10
geocode many addresses.. – ArcGIS North America Street Map
• Example: http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//001z00
000039000000.htm
– Employee residential data of a – The best street database that the library has. A variety of pre-set geocoding options are
company A, located in Chicago included, such as street alias search.
– Available to the NU affiliated members, contact Chieko, c-maene@northwestern.edu
– Total cases (N) = 24,158 – Free 1-hour Online Seminar: Geocoding in ArcGIS Desktop 10
http://training.esri.com/acb2000/showdetl.cfm?did=6&Product_id=986
• Tip to increase performance: – Tutorial
– Clean your data first http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#/About_t
he_Geocoding_tutorial/002500000006000000/
• Remove POBox addresses
– Tutorial in PDF
– Pre-sort your input by zone field http://help.arcgis.com/en/arcgisdesktop/10.0/pdf/geocoding-
• Sort by state, city, zipcode tutorial.pdf

Other Geocoding Tools Other Geocoding Tools (cont.)

• ArcGIS is not the only option – Batch Geocoder
we have! – http://www.batchgeocode.com
– ESRI Address Coder – Check output accuracy – I had some problems
– http://www.esri.com/softwar – GPS Visualizer
e/coder/ – http://www.gpsvisualizer.com/geocoder
– Stand-alone geocoding tool – Similar to batch geocoder
– Expensive: $9,000 + – Also, visit this near complete list of geocoding tools!
– Easily add extra data https://webgis.usc.edu/Services/Geocode/GeocoderLi
{demographics/marketing st.aspx
data} to your geocoded
points