General guidelines for optimizing geocoding – Pitney Bowes MapMarker USA User Manual
Page 55
Chapter 3:
Understanding the Geocoding Process
MapMarker USA 25
55
User Guide
There are trade-offs that you must consider with respect to how you intend to use the geocoded
data. When setting up a geocoding session, consider questions such as the following:
•
What level of matching accuracy are you looking for: a unique address match or close match?
•
What level of geographic accuracy do you need for your geocoded points: street-level centroid or
postal code centroid?
•
Is your goal to geocode as many records as possible?
The answers to these questions are driven by how you intend to use the geocoded records. When
determining the location of a new retail store, you need to know the distribution of current and
potential customers. In this case, you would need to geocode as much of the customer base as
possible but you might not need an exact street address match for each customer. Geocoding to
postal code centroid would be suitable for this analysis.
Street level geocoding with strict matching preferences would be useful for cases in which it is
necessary to know exactly where your customers are. For example, if you are a utility service
coordinator and need to know exactly where your customers are in relation to neighborhood gas
lines, the positional accuracy of each customer is of critical importance to you.
In addition, the matching routine uses the geocoding preferences that are set for the table to
determine whether certain matching conditions are required or not. For example, the MapMarker
default preferences do not require a match on street name and ZIP Code, but require an exact
match on the house number. This generally gives the best combination of hit rate with the fewest
erroneous matches (false positives) and the best performance.
For an overview of geocoding match strategies, see
General Guidelines for Optimizing Geocoding
MapMarker USA 25 has been optimized for high performance. Follow these guidelines to achieve
even faster performance.
In general, you will achieve faster and more accurate geocoding if you standardize and format your
data. By reducing the ambiguity in the input data, MapMarker USA 25 will be able to find appropriate
candidates faster. For complete information on addressing practices and conventions and ZIP
Codes, see th
age on the USPS web site.
Also consider the following guidelines to optimize your results and performance.
•
Use the fastest processor available to you. Pitney Bowes Software recommends 2.5 Ghz dual
processor or better.
•
Have enough memory so that the operating system can allocate some memory to your disk
cache. Pitney Bowes Software recommends 2 GB RAM or better.
•
Have at least 10 GB available disk space.
•
Store the Address Dictionary on a local hard drive.
•
Sort your table by postal code before geocoding.
•
Use Must Match (exact match) criteria for the most precise geocoding results. However, note
that this strategy requires more processing and may reduce performance.
•
Do not create points automatically.
•
Remove indexes on any table output columns before geocoding.