ArcMap Address Matching Lab

Due in two weeks

Objective: The objective of this lab is to expose you to another of ArcMap's capabilities: address matching, sometimes called geocoding. The notion is to convert street addresses (like 231 Nanana Street) to coordinates. This is done by matching the addresses against a database of street segment records, conceptually like those in GBF/DIME, to find the data record with the best matching street name and address range, and then interpolate along the segment to associate a geometric point with the address. This functionality is provided in the Geocoding Tool and seems to require the use of a (personal) geodatabase.

In this lab you will:

1. Create an "Address Locator" (arc-speak for making an indexed database out of a street segment shapefile for address matching),
2. match an existing file of addresses against that "Address Locator" to produce a shapefile from the list of addresses,
3. plot the matched addresses on a map and examine the unmatched addresses to gain insight into what constitutes 'problem addresses'.
4. practice reading the 'help' in ArcGIS.

Data Needed

The data have been placed on laulima, emailed to you, and probably in the \\Odin\Data\Geog_488 server directory. You may already have that drive "mapped" or the data copied to your own disk.

• allroad3 shapefile (in all its parts) - the Oahu street segment data derived from Census and USGS DLG data, it has left and right address ranges and zipcodes. (The C&C Street data don't work as they don't have these left and right address range fields.)
• restaura.txt - a text file list of restaurants extracted from a recent yellow-pages

Game Plan

1. In ArcMap's and/or ArcCatalog's Help function(s), read about Geocoding and Addresses, and particularly about the address locator until you have a feel for how Arc is going to do this. The material indexed under "addresses, locator style" and "addressess, matching" (the latter is an external .pdf file that tells about the rule base for address matching) is especially useful.
2. Create your "address locator" from the allroad3 shapefile by following the directions in the "Creating an address locator" help page. Basically, in a geodatabase, save your allroads3 data and add a new locator via ArcCatalog. The address locator style "US Address Dual Ranges" works fairly well, then in the Field Map, see that "FromLeft" = LEFTADD1, "ToLeft"=LEFTADD2, "FromRight"=RGTADD1, "ToRight"=RGTADD2, and "Street Name" = FNAME. You might experiment with other styles.
3. In ArcMap, geocode the addresses by following the directions in the "Geocoding a table of addresses" help. (Basically, add the addresses to the project and select Geocoding from among the "Geocoding" Tools. Select the "address locator" and the table to geocode. Indicate which field holds the address data and where to save the result (U:\Geocoding_Result_1 for instance). Run it, and it should add the output to your map on screen.
4. Produce the deliverables and answer the questions below.

Hand-in:

1. A map showing the distribution of matched restaurants.
2. A table showing the counts of the matched, the unmatched, and the partially matched addresses (summarize on status) for each of the "address styles"
3. A one page write up indicating:
   • what address information was in the file that you geocoded,
   • which "styles" of address locators would best fit these data, which of those you tried, and which of those seemed to work best with these data?
   • An assessment of the apparent kinds of problems the software had in matching the unmatched addresses.
   • Your observations on what might make this work better.

Hints

Help - is on the main menu and on-line. You will be better off getting used to using it to help solve problems than not doing so.

Making new "address locators" takes only a few seconds and greatly affects the results. In one recent class test case, just switching from "US Streets" to "US hyphenated ranges" increased the match rate by more than 50%.