Geocoding Techniques and Options for US and International Locations Brady Hoak, Tosia Shall
Agenda What is geocoding? Requirements for Geocoding Preparing Your Data Selecting a Locator Geocoding Process Additional Geocoding Topics
What is Geocoding? Process of assigning a geographic location (X,Y) to a coordinate, address or the description of a physical feature - 38.45 N, 123.05 W - 380 New York Street - Well Head BF-245 Wide range of applications including business analytics, logistics, spatial analysis
Requirements for Geocoding
Requirements for Geocoding An understanding of the process Data to be geocoded Locator(s)
Requirements for Geocoding An Understanding of the Geocoding Process: Prepare your addresses Build or obtain reference data Determine address locator style Build or use an address locator Locate addresses, rematch remaining addresses
Requirements for Geocoding Data to be Geocoded: A Table of : Addresses - 380 New York Street, Redlands, CA 92373 Place Names Mount Rushmore Descriptive Locations - Well Head BF-245 A Single Location: Address 8615 Westwood Center Drive, Vienna, VA 22182 Place Name Grand Canyon Descriptive Location - 50 Feet NW of 10403 Main Street, Fairfax, VA 22030
Requirements for Geocoding Locators: ArcGIS Online (Services) North America, Europe, World Gazetteer ArcGIS Server Service Publish your own geocoding service with ArcGIS Server Local Locator Choose an existing locator on your machine Build your own locator to use locally
What is a Locator? Data, rules and indexes that are used for geocoding Locators available as a dataset Can be used locally or as web services Locator Streets, address points, postal points Address rules and conventions Address Information Rules Reference data Indexes Map Features
New Geocoding Engine for ArcGIS 10 Better Quality match rates, explicit scoring International data support including Unicode Accurate, Configurable, Extensible - Single line inputs are standard - Improved results (+1% to +10%) - Poorly formatted data better handled Supports legacy locator use 123 Global Street
Overview of the geocoding engine at ArcGIS 10 Geocoding Process Grammar Data structure Aliases Scoring
Geocoding process Unstructured Input 100 fifth avenue ny 100 5th Ave, New York NY 10011
Geocoding process Unstructured input 100 fifth avenue ny House PreDir PreType Name SufType SufDir City 400 N New York St Redlands 100 5th Ave New York Structured Reference Data
Geocoding engine : Grammar All supported forms of addresses explicitly defined in a grammar Grammar is determined by the local culture - Language(s) - Order(s) of elements - Variability of elements
Geocoding engine : Grammar (continued) Grammar example: Address: House StreetName City // 380 New York St Redlands StreetName & StreetName City // Main St & 2nd Ave Springfield SpatialOperator Address ; // 100 ft SW from 3140 Canyon St Carlsbad StreetName: PreDir PreType Name SufType SufDir ; House: number // 123 number - number // 17-100 number letter ; // 100A PreDir: N E W S NW SW NE SE ; PreType: Ave Hwy ; SufType: Ave St Rd ; Handles ambiguous addresses No standardization
Geocoding Engine : Reference Data Structure Locator stores a snapshot of reference data List of unique values for each reference data field Many-to-many relationships between fields House PreDir PreType Name SufType SufDir City 11 72 100 302 400 520 1000 N E S W Ave 1st 2nd 5th Alabama Alpine Bacon Brockton Cimarron E New York Zzyzx Ave Ct Ln Pkwy Rd St N NE E SE S W Amarillo Redlands San Diego New York Zion 72 N Alabama St, Amarillo
Geocoding Engine : Reference Data Structure Some of the fields are used for search Name City 1st 2nd 5th Alabama Alpine Bacon Brockton Cimarron E New York Zzyzx Amarillo Redlands San Diego New York Zion
Geocoding Engine : Reference Data Structure Reference Data is an extension of the grammar SufType: Ave St Rd Ct ; Ln Pkwy ; PreDir PreType Name SufType SufDir City N E S W Ave 1st 2nd 5th Alabama Alpine Bacon Brockton Cimarron E New York Zzyzx Ave Rd St N NE E SE S W Amarillo Redlands San Diego New York Zion Reference data helps dealing with some local addressing oddities without the need for locator customization
Geocoding engine : Aliases Common abbreviated forms (aliases) - Mt, Mtn, Mount, Mountain mean the same in the context of a City name - Same for View vs. Vw - There are many ways to spell Mountain View - Mountain View - Mt View - Mtn View - Mount. View - Mountain Vw - Mt Vw - Mtn Vw - Mount. Vw
Geocoding engine : Scoring Grammar has information about relative importance of the components Candidates with same best score are considered ties only if their geometries are different More precise score (in decimal number) Score computation and weights are configurable Address: House 2 StreetName 6 City 3 House 2 StreetName 6... StreetName: PreDir 5 PreType 6 Name 70 SufType 7 SufDir 5 ; There is no penalty if some of the components (i.e. City, ZIP) are missing from the input address
Geocoding engine : Scoring (continued) Address: 100 Fifth Avenue NY (2 * 1 + 6 * 0.95 + 3 * 1) / (2 + 6 + 3) = 0.97 House: 100 StreetName: Fifth Avenue City: NY 1 (5 * 0 + 6 * 1 + 70 * 1 + 7 * 1 + 5 * 1) / (5 + 6 + 70 + 7 + 5) = 0.95 PreDir: PreType: Name: Fifth SufType: Avenue SufDir: 1 (alias applied) 0 1 1 (alias applied) 1 (alias applied) 1 House PreDir PreType Name SufType SufDir City 100 N 5th Ave New York
ArcGIS 10 Geocoding Engine Summary New engine from the ground up Handling of ambiguous addresses - Example: (10 West Ridge Drive) No penalty for missing zone information Can parse any input field, not just Street Name Returns matching address as it is written in the reference data Single line input Highly configurable - Single XML file
Preparing Your Data
Best Practices for Preparing your Data Check for errors - Spelling - Numeric address ranges - Missing information Tables can be: - Geodatabase tables - Comma or tab delimited.txt - Excel Spreadsheets - Field Names in CAPS, define Field Type in Excel - Microsoft Access tables - Other data sources accessed through OLE DB (connect to an RDBMS, for example) - dbase
Selecting a Locator
Access Locators By Using an ArcGIS Online geocoding services Using an ArcGIS Server geocoding service Using a local locator when using Desktop
Using an ArcGIS Online Geocoding Service Geocoding at the Address Point and Address Range level - North America & Europe locators - Updated semi-annually (quarterly coming, NA) - Designed for small / medium volume of addresses World Gazetteer Free and for subscription
Using an ArcGIS Server Geocoding Service Use ArcGIS Server to publish your own Geocoding Service - You can provide this service to your entire organization internally - Addresses to be geocoded are not sent through the open Internet You will maintain the Locator - You must update and maintain the reference data - You will rebuild the locator after making any reference data changes
Where else can I find Locators to use? ESRI Data & Maps - In-the-Box! - Free for all users with ArcGIS - 2005 data from TeleAtlas - For large North American datasets ESRI StreetMap Premium - Geocoding at Address Range and Address Point Level - Available for North America & Europe - Leveraging data from Navteq & TeleAtlas - Updated semi-annually Create your own
Geocoding Process
Geocoding Process Choose a Method Batch Reverse Interactive
Interactive Single Line Desktop Geocoding Use the Geocoding Toolbar: Choose a Locator Type an address
Interactive Desktop Geocoding - Find Tool Single Line Input at ArcGIS 10
Interactive Single Line ArcGIS Online Geocoding
Single Line Geocoding in a Web App
Interactive Desktop Geocoding Re-matching
Interactive Geocoding Demo Desktop ArcGIS Online Tosia Shall
Batch Geocoding Using the Geocoding Toolbar in ArcMap Using the Geocoding Tools in ArcToolbox Using a Web App configured with a Geocoding Service
Using the Geocoding Toolbar in Desktop In the Geocoding Toolbar: Use Geocode Addresses Tool for Batch Geocoding Choose a Locator
Geocode Addresses Tool (Geocoding Toolbar)
Using the Geocode Addresses Tool in ArcToolbox Use Geocode Addresses Tool
Batch Geocoding in ArcGIS Explorer Online
Batch Geocoding Using a Web App
International Locators ArcGIS Online Map Viewer Search - Optimized for Single Line Entry - Many countries, many continents - Different from standard and premium tasks available on ArcGIS Online
International Locators Standard Tasks tasks.arcgisonline.com Support single address, reverse geocoding and batch geocoding - Up to 1,000 batch geocodes in a year - Batch geocode up to 10 addresses per request Connect directly to these online tasks through ArcGIS.com, the ArcGIS APIs, and ArcGIS 9.2 or higher Esri Places World TA_Address_NA TA_Address_NA_10 TA_Address_EU TA_Streets_US TA_Streets_US_10 - Address locators based on NAVTEQ 2010 Q4 reference data for North America - NAVTEQ 2011 Q1 reference data for Europe - Street locator based on NAVTEQ 2010 Q4 reference data for US - Place finder based primarily on the May 2011 GeoNames data See help.arcgis.com - ArcGIS Online Content Help
International Locators Premium Tasks premiumtasks.arcgisonline.com/desktop/services Support single address, reverse geocoding and batch geocoding - Purchase in blocks of 25,000 - Batch geocode up to 1000 addresses per request Activate subscription, follow instructions found in ArcGIS Online Content Help Using the World Geocoding Subscription Esri Places World TA_Address_EU TA_Address_NA TA_Streets_US TA_Address_NA_10 TA_Streets_US_10 TA_BatchAddress_EU TA_BatchAddress_NA TA_BatchAddress_NA_10 - Address locators based on NAVTEQ 2010 Q4 reference data for North America - NAVTEQ 2011 Q1 reference data for Europe - Street locator based on NAVTEQ 2010 Q4 reference data for US - Place finder based primarily on the Dec 2011 GeoNames data
European Locators Premium Tasks The locator is based on NAVTEQ Q1 2011 reference data for Europe and is limited to the coverage provided by NAVTEQ: Andorra, Austria, Belgium, Denmark, Finland, France, Germany, Gibraltar, Ireland, Italy, Liechtenstein, Luxembourg, Monaco, The Netherlands, Norway, Portugal, San Marino, Spain, Sweden, Switzerland, United Kingdom, and Vatican City
Batch Geocoding International Locators Tosia Shall
Additional Geocoding Topics Brady Hoak
Composite Locator Combines multiple locators - Geocode against multiple datasets - Cascade/Fallback geocoding - Spatially disjointed datasets
Geocoding Resource Center http://resources.arcgis.com/content/geocoding/10.0/about
Help Resource