Geocoding:
A practical definition The
process of placing data on a map.
Before
geocoding
Name
Address
Zip
Bill Smith
123 Orca St
98000
After
geocoding
Geocoding:
A technical definition
The process of matching records in
two databases: your address database (without map position information)
and a reference street map or other "address dictionary".
Geocoding software (e.g., MapInfo Pro or MapMarker)
links records in the two databases by matching street names and
address numbers: e.g., matches the street name in the address to
a street name in the street map etc.
When your database records are successfully linked
or matched to a
reference street map database, your records are tagged with the correct map positions,
typically lat-lon coordinates.
Thereafter, your data table carries its own position
information and can be mapped without using the reference street map or address dictionary.
MapInfo Pro, by default, hides the map coordinates:
i.e., it does not list the Lat-Lon values in the tables. But it uses
the coordinates, nevertheless, behind the scenes.
You can map addresses by zip code, of course,
without geocoding at the street level. This produces a very coarse
understanding compared to a address-level point map because you
consider all those addresses to be in the same place: the zip code.
After geocoding by
zip code
After geocoding by address
To summarize data by ring
buffer or any custom shape
Which district office should handle the new
client? Geocode client addresses by street address and see which
territory region they fall within.
How many customers fall with 5 minute drive of
each proposed espresso outlet? Geocode customers by street
address, then overlay a 5-minute drivetime polygon.
Are capital improvement projects benefiting all
legislative districts equally? Geocoding the street addresses of the
capital projects then overlaying the legislative districts.
Banks and insurance companies often report
statistics by census tract. To determine the census tract for a loan
or depositor, geocode the street addresses, overlay the tract map,
then use Query > SQL Select to summarize by tract.
To target more precisely
We may make assumptions about households
based on the characteristics of the neighborhoods where they live.
If geocoded by street address, we can associate household records with
smaller, more homogenous areas such as block groups. This
results in more accuracy.
As shown below, the demographics of the block
group areas (smaller) can be dramatically different than the
demographics of zip codes (larger).
Higher income zip codes
Higher income block groups
Geocoding: Part of modern corporate
customer information systems
More and more Fortune 1000 corporations are
geocoding all their customer addresses as a standard practice. By
making map position part of their overall corporate information system
design, all areas of the organization take advantage of spatial analysis
techniques.
The original way to geocode used MapInfo's built-in
geocoding engine along with a street map. MapInfo Corp pioneered
this technique for PCs, releasing MapInfo (for DOS) in 1987.
MapInfo for Windows, released in 1992, also offered street level
geocoding using displayable and user-editable street maps.
Geocoding is accessible via the MapInfo menu option
Table > Geocode ... Just add street map data.
Requires a street map with street names & address
ranges: e.g., "StreetPro", "StreetInfo" or comparable.
Note: The most common version of StreetPro is "StreetPro
Display", which does not have the address number ranges.
Uses an "abbreviation file" (e.g., mapinfow.abb)
to standardize some address components: e.g., "Street", "Str", and
"St".
Can use a City or Zip code boundary map to distinguish
addresses that occur in more than one place in a county.
Advantages of MapInfo Pro
Cheaper than MapMarker if & only if all addresses are in
a few counties at most.
Street address information in street tables is completely
editable. Missing a street? Just draw it and fill in the
street name and address fields. You can match against this new
street immediately.
Can easily use new street map information generated by local
governments.
Suitable for use in Mexico & other foreign countries for which MapMarker
versions may not be available.
MapMarker: Single-purpose
stand-alone geocoder
A standalone geocoder software program,
introduced by MapInfo Corp. in 1996 and now at Revision 12.1 (January 2007). It
can run without MapInfo Pro and is optimized to do one thing: geocode
large datasets quickly -- including data set covering many states.
Primarily a regularly-updated data
product, but includes a simple menu and dialog UI.
Available in multiple variants, each with different pricing and
different levels of precision. The most common version is
MapMarker Plus. Other versions include MapMarker Zip4, MapMarker
Std, MapMarker Plus. There are also database specific extensions
for SQL Server and Oracle.
Advantages of MapMarker
Cheaper, except for one-two-three county applications.
One-step geocoding, even if addresses are in many different
counties across the US.
Cleans or standardizes addresses, zip codes, zip+4 codes on
the fly. Can produce USPS "CASS-certified" mailing lists.
[CASS-certified means lower postage rates for bulk mailers.]
Includes Zip+4 reference files and can automatically compute
the Zip+4 code and geocode to the Zip+4 location as a fallback, if it cannot match to exact street
address.
Very fast: SGSI has processed 1,000,000 records in under
1
hour in batch mode.
Adds census tract, block group, and block codes to the
address record as part of its single-pass processing. No separate SQL processing
required.
Friendly, straightforward user-interface. Easy to use.
Is engineered as a true client-server application and
can be used as the geocoding engine for a web server, for example.
More sophisticated "rule based" matching
algorithms are better than those in MapInfo Pro. Offers a better set of
"close-match" candidates than MapInfo.
Can process sets of address files, running in
"batch" mode.
While address and street data is not directly
user-editable, power users can build & compile custom MapMarker-format
databases and use them in concert with MapMarker's standard databases.
Server version is 100% Java, with Java and XML
api's for integration with custom applications.
MapInfo Pro + MapMarker: Best
of both worlds
With MapInfo Pro 8.5 and MapMarker, users geocode MapInfo tables or
selected records without leaving MapInfo Pro.
Improved geocoding precision is possible with better underlying data and
better software algorithms.
Since all geocoders rely on digital maps, geocoding
becomes more precise as the underlying street maps, street names & street
number content is improved.
There is a recent trend though, toward basing address
geocoding improvements on entirely new digital map layers--not just street
maps.
Address Interpolation
Most geocoding engines rely on house number
interpolation to estimate the position of an exact street address.
This technique is excellent where the house number range
reflects actual house numbers and where homes are evenly spaced along the
block.
It can create a distorted picture, however, if the
actual addresses along the block aren't evenly spaced or if the actual
high-low house numbers are very different than the theoretical high-low
house numbers.
This is the approach taken by most commercial geocoders.
E.g.,
Enhanced Address Interpolation
With enhanced interpolation, the geocoding system
computes the *actual* high-low house numbers on the block, by reference to
extra data beyond the street map. Interpolation still occurs, but it
is between more accurate high-low locations and the results are much better.
This is the approach pioneered by MapInfo beginning with
MapMarker Revision 12.0:
Parcel-level precision
Of course, enhanced interpolation solves only part of
the problem. In many cases, homes are *not* evenly spaced along the
block.
To resolve this problem, MapInfo has incorporated actual
land parcel boundaries as an extra-cost option for MapMarker. Users
may specify that addresses be geocoded to either the centroid of the parcel
polygon or to a point aligned with the parcel but offset a specific distance
from the street center line.
Note: This Parcel-level option does
not include a visible parcel map layer. Also, while millions of
parcels are referenced, the coverage is not 100%. Please contact SGSI
at 206-224-0800 for details.
Software + Your data: Custom
dictionaries
The next step toward maximum geocoding precision is
determining x,y coordinates based on the most important part of the
property. E.g., a delivery service may want to determine the lat-lon
of the doorway.
The above prices are ballparks only and cover desktop
use (not server) for 1 named user. Substantial discounts apply for
3-Year Update Subscriptions. Modest discounts apply for government
users. Please
contact SGSI at 1-206-224-0800 or email us
for more specific pricing.
