Course Materials for Geography 481, Project Eight, Geocoding

In the course thus far, the data for most of the exercises have been in a form that is readable by ArcMap. However, there exists a vast quantity of useful information that needs some manipulation before it can be utilized by ArcMap. One type of spatial data that we use frequently is the street address (e.g., 1234 S. Main St.). Yes, this is spatially referenced data, but ArcMap does not understand this designation without some help. ArcMap prefers some sort of x,y coordinate for spatial data. Wouldn't it be nice if there was some way to automatically convert street addresses to x,y coordinates? Well, there is... sort of. The procedure is called address matching or geocoding. I say "sort of" because geocoding is fraught with complications as you will see.

In this project we will create a new layer (and shapefile) of points generated from their street addresses. Geocoding requires a pre-existing reference layer (usually streets) to serve as the base map and an address table containing the addresses you want to locate as points on a map.

Your task in this project is to prepare a layer to be geocoded and address match an existing file of hazardous materials locations.

Setup

The Streets layer will be our reference layer.

Now we need some addresses. We will be batch geocoding a list of addresses from a table/file. Although we have been using DBF files in class thus far, ArcMap can use Excel files, though with limited capability (and occasional "excitement"). Add the hazmat.xls as follows:

Click Add Data
Double-click hazmat.xls
Click hazmat$ and click Add

You have now added a table of businesses with their address information. Open it and take a look. Can you find the address information? How many different columns contain parts of the address for each record?

Close the table when done.

Building an Address Locator

An address locator is part of the larger geocoding framework. According to ESRI, the address locator "defines the techniques to be used by the geocoding engine in interpreting addresses against the rule base files." There are many variables and parameters that must be set for successful geocoding. The address locator is where these settings are maintained. Address locators are built in ArcCatalog.

Before we begin, we need to have a look at our reference layer. Open the attributes of streets table. Look at the different fields describing each street segment. The fields L_f_add and L_t_add contain the lowest and highest addresses on the left hand side of the street segment. Surprise! The fields R_f_add and R_t_add contain the lowest and highest addresses on the right hand side of the street segment. Close the table.

Now that we know how the data are formatted in the reference layer (streets) and the table we wish to geocode (hazmat.xls), we are ready to build our address locator.

Open ArcCatalog by clicking the ArcCatalog tab on the right of your screen (or the ArcCatalog button)
Click the + next to Folder Connections
Right-click c:\temp (if missing create the c:\temp\ connection)
Click New then Address Locator to open the Create Address Locator dialog
Click the yellow folder icon next to Address Locator Style and choose US Address - Dual Ranges since we have ranges for both sides of each street. You will get an error message until you have selected Reference Data.
Select Streets for your Reference Data
Scroll down, inspecting the Field Map along the way. ArcMap automatically maps to fields it recognizes.
Save your Output Address Locator as c:\temp\happy
Click OK... this may take a few seconds to process. Observe progress in the lower right of the screen or in the dialog window.

The address locator you created has a set of rules that makes the streets layer address matchable.

Address Matching Process

Proceed slowly and pay attention.

Voila! The addresses from the hazmat.xls have now been given x,y coordinates. Well, most of them anyway.

Re-Matching & Geocoding Preferences

The matching process does not always hit every single address. Typographical errors, missing addresses or other differences may not allow a complete match when geocoding. Interactive re-matching allows you to correct for some of these errors.

Click the Customize menu in ArcMap, then Toolbars, then Geocoding to open the Geocoding toolbar
Select your happy address locator in the left-hand drop-down.
Highlight the haz_geo layer in the TOC
Click the Review/Rematch Address button on the Geocoding toolbar (last mailbox on the right. Will be grayed out unless you select the layer in the TOC)
In the Show Results drop down menu select Unmatched Addresses.
The top window displays the addresses in hazmat.xls that did not successfully geocode.
The bottom portion of the dialog allows you to make modifications so that the address matches.
In the bottom middle window, the left column displays the "score" the address received when matching. The score is based on the address match with spelling, street type and direction. A perfect score is 100 and anything below set thresholds will not geocode. It may also display "candidates" for the selected record that were close, but did not meet the minimum match score.
If the list of potential matches contains the street range you think is correct, make sure your choice is highlighted and click Match.
If the list is empty or does not contain a suitable candidate, look at the address to see if you can identify a possible problem, usually it is the misspelling of the address. Note: Misspelled street names can occur in the street reference file as well as in the address database file.
Click the Geocoding Options button in the Interactive Rematch dialog to open the options dialog.
If a spelling error is suspected, lower the minimum spelling sensitivity by ten or so points and click OK to apply the changes. Hint: If you know it is a misspelled name, you can type the correct spelling directly into the text box called Street or Intersection.
Other potential errors may have occurred in how ArcGIS standardized the address. If it is reading 800 St College as street number = 800, street name = college, street type = St, then obviously it has standardized the address incorrectly.

In this section you will interactively match the three addresses that were un-matched by doing the following:

Juniper is an incomplete address (it is missing ST NE)

In the top window of the Interactive Rematch dialog, scroll to the right to the Name and Address fields
Verify that the Juniper address is selected in the top window
Notice there is a Candidate match listed in the bottom middle window that includes the missing ST NE. Its score is less than the minimum, so it did not get geocoded, but it is the correct match.
Click on the listed Candidate and then click the Match button at the bottom, then click Refresh at the top
The Juniper record should have been matched and disappeared from the list at the top

10th ST NE has an incorrect zip code (it should be 30309)

In the top window, scroll to the right to the Name and Address fields
Verify that the 10th ST NE record is selected in the top window
Notice there are no Candidate matches listed in the bottom window
In the bottom left window, change the ZIP Code to 30309, then click Search at the bottom
Candidates are now listed in the bottom window, one for each side of the street (555, 556)
Click the 555 candidate, then click Match at the bottom, then click Refresh at the top
The 10th ST SE record should have been matched and disappeared from the list at the top

Mirtel is a misspelling (it should be Myrtle)

In the top window, scroll to the right to the Name and Address fields
Verify that the Mirtel record is selected in the top window
Notice that again there are no Candidate matches listed in the bottom window
In the bottom left window, change the name listed in Street or Intersection to Myrtle, then click Search at the bottom
Candidates are now listed in the bottom window, one for each side of the street (844, 845)
Click the 844 candidate, then click Match at the bottom, then click Refresh at the top
The Mirtel record should have been matched and disappeared from the list at the top

Now you should have finished geocoding all of your addresses.

If no more records are listed in the top window, Close the Interactive Rematch dialog.

Many of these modifications can be adjusted before geocoding. This may be appropriate if the address database contains many spelling errors or the street directionals, suffixes or prefixes are missing. Unfortunately, this can be a common problem and very time consuming to fix.

Interactive Geocoding (without creating a layer)

If you only want to find a single address and do not need to create a new layer, you can bypass the matching process and use the Find Tool on the menu (it looks like binoculars) to find the address on the map. There must be a a valid address locator for this to work.

ArcMap places a point at the address location. This point is a graphic marker, just as if you had used the Add Graphic tool. It is not part of a layer (shapefile or feature class). If you want to create a point layer containing address locations and attributes of those locations, you should use the geocoding approach that you used earlier in this exercise.

On Your Own

Assume your house is located at 742 Durante Pl NE in the 30308 zip code. If I tell you that the data coordinates are decimal degrees of latitude and longitude, can you identify all facilities with hazardous materials located within one-quarter mile of your house?

Please produce a printed (PDF) map meeting the above criteria. All cartographic conventions apply.

Last modified 11/09/2021