Address/Spatial

Overview

The Address/Spatial group includes tools for performing address hygiene operations, spatial processing, and geocoding.

Address matching

Real-world address data tends to be non-standard and "dirty," and attempts to match the raw address text often produce poor results:

• Some components of the address are more critical than others and should have a tighter match criteria. For example, the house number is almost always more critical than the street suffix or directionals.

• Common variants in address entry tend to turn similar addresses into very different strings, for example E HWY 20 and US HIGHWAY 20 EAST.

• Most address matching software first standardizes the addresses, then matches on the standardized address. This process is flawed for several reasons:

• Some street addresses (especially rural addresses) are not USPS deliverable, and standardization software fails.

• Some addresses are so "dirty" that standardization cannot correct the errors.

• Address standardization tools are generally more expensive and slower than Data Management.

Data Management takes a different approach. Instead of trying to create a single standardized address, it recognizes the inherent ambiguity in dirty address data, and generates multiple candidates for each address. Each candidate represents a slightly different interpretation of the address. The Parse Address tool applies standardization to basic address elements such as street suffixes and unit designators to improve matching. And it breaks the address down into components such as house number, predirectional, and street name, so that you can apply different weights and matching functions to each component.

Using the Parse Address tool, you expand each input address into one or more candidate parsed addresses, and then match the sets of parsed addresses against each other. Because each candidate is matched independently, you greatly improve your odds of finding matches in dirty data.

In this section: