27

Is there a good updated source for comparing existing Geocoding APIs (i.e. Google, MapQuest) and their features?

I'm really interested in their language support as well. For example, Google has libraries for its Geocoding service in JavaScript, Java, Python, and many more.

Taras
  • 32,823
  • 4
  • 66
  • 137
TomSchober
  • 373
  • 1
  • 3
  • 6

7 Answers7

27

I did a simple comparison a year ago showing eleven different geocoding services, free ones as well as pay services, and the results are in a google spreadsheet. I work at SmartyStreets, so you'll see that listed in the first column but not in first place. I tried to make the comparison unbiased so the results are actually useful. I have now opened the comparison chart for editing by anyone. If you can add to it, please do. Keep your edits objective, don't try to spin things.

geocoding comparison chart

The exact location was determined visually by finding the location on googlemaps and dropping a pin there to obtain the lat/lon.

Jeffrey
  • 1,484
  • 12
  • 13
  • How did you find the exact location of those properties? Was it GPS around your area? Why is Bing empty? Interesting data for sure. – Michalis Avraam May 31 '13 at 16:25
  • 1
    The exact location was determined visually by finding the location on googlemaps and dropping a pin there to obtain the lat/lon. When I originally did this, I didn't even think of Bing, though I am adding it now. I'm adding more data points now. I actually have a list of all the McDonalds in the United States with confirmed lat/lon if someone wants to help make a more comprehensive chart. – Jeffrey May 31 '13 at 16:39
  • You can use geopy to quickly geocode with a lot of those providers. It probably wouldn't be too hard to add a SmartyStreets provider there too - it probably won't hurt users to know they have access to that too. – Michalis Avraam May 31 '13 at 17:21
  • That's a nice list of available APIs and the accuracy of some geocoding. So helpful, but what I'm really looking for is an existing comparison of the FEATURES (i.e. has a rest service, has a SOAP service, has a JAVA api, javaScript API, python API, etc) – TomSchober Jun 03 '13 at 16:20
  • 1
    Looks like I need to add that. Give me a little time. – Jeffrey Jun 03 '13 at 18:43
  • 2
    I've added a second sheet to that comparison. The first sheet compares the geocoding precision and the second sheet compares the features and services offered. – Jeffrey Jun 03 '13 at 20:23
  • 1
    @TomSchober if you will come up with a complete list of what you'd like to see between the services, I'd be happy to fill in the details. – Jeffrey Jun 03 '13 at 20:24
  • 1
    Jeffery. I think it would be a fun project to build a comparison tool from your McDonald's list, assuming you would still be willing to share that. You can get ahold of me through http://www.fredtrotter.com/contact-fred-trotter/ – ftrotter Dec 26 '14 at 06:12
  • @ftrotter - I just emailed you a list of all the McDonald's, Wendy's, and Wal-marts. – Jeffrey Dec 26 '14 at 17:09
  • It's an interesting comparison. I'd just ask why the differences between Nominatim and OpenStreetMaps. Aren't they the same service? – Gustavo Meira Mar 23 '15 at 13:59
  • 1
    @GustavoMeira Under the head they should both be the same service yielding the same results, but in my testing they were different. Likely it's a different implementation of the geocoding algorithm. – Jeffrey Feb 29 '16 at 17:14
4

There is a free REST API at http://matchbox.io. It standardizes the address and returns lat long as well as other information. For example:

http://api.matchbox.io/address/?addr_1=1390%20Market%20St%20%23107%2C%20San%20Francisco%20CA%2094102

Joel
  • 141
  • 2
  • Thought it appropriate to add that the Quick Guide at the time of this comment states this concerning Rate-Limit: The Address API is free to use for up to 60 requests per minute – Jared Eitnier Dec 31 '15 at 04:13
3

I don't know of any such comparison, but I would like to point out that Bing has a REST endpoint you could access from any platform.

Russell at ISC
  • 1,912
  • 10
  • 11
  • I have added just such a comparison to my geocoding comparison gdoc. (https://docs.google.com/spreadsheet/ccc?key=0AidEWya_p6XFdGw1RmZ6TjB1ajZxVk81d2pISDMzVUE&usp=sharing) Feel free to add to it you have information. – Jeffrey Jun 10 '13 at 17:49
3

The Texas A&M Geoservices site has a comparison of the main features of several geocoding systems - http://geoservices.tamu.edu/Services/Geocode/About/GeocoderList.aspx. We try to keep it as up to date as possible.

dwgoldbe
  • 39
  • 1
2

I have found this http://www.gisgraphy.com/compare/ with Gisgraphy, Nominatim, Google Maps, Yahoo! Placefinder and MapQuest compare.

Fezter
  • 21,867
  • 11
  • 68
  • 123
aborruso
  • 1,279
  • 10
  • 21
0

Rather late, but we at NetToolKit have invested a lot to build a geocoding comparison tool (released late 2018). The tool is primarily oriented around comparing accuracy of results (you can see how far one result is from other results by clicking on it and checking the legend), but the table below also highlights cost differences and differences in terms of service. We compare several providers, including OpenStreetMap/Nominatim, PostGIS, and the US Census geocoder. If you have any feedback, please let us know.

NetToolKit
  • 11
  • 1
0

My company https://bettermile.com/ does benchmark multiple geocoders on a consistent basis with millions of addresses we receive from parcel delivery companies in Europe, mostly Germany, Italy, Spain, Austria, the Netherlands and more recently Canada.

Here’s a sneak peek of what we have seen so far with Google, Here and Mapbox. I also included our product because why not :), but the results look like this.

Accuracy Level BetterPlaces Google Here Mapbox
HIGH 97.90% 97.65% 96.89% 69.30%
MEDIUM 1.02% 1.37% 1.96% 27.97%
LOW 0.34% 0.98% 1.10% 2.73%

The accuracy levels mean:

HIGH: The result is accurate to the house number, the building (rooftop coordinates) or to a Point of Interest level.

MEDIUM: Accurate to the closest street or in front of the building

LOW: City, town or locality

There are a few of things to mention:

  • These are averages of all the multiple benchmarks we have done in the last 6 months or so, all of them with millions of addresses, and all of them across different area types (urban and rural)
  • We apply normalization to the addresses before sending to the different geocoders, which means that a lot of corrections have been made and hence the quality of the result and the success rate is significantly higher.
  • Without normalization you can discount at least 5%-8% from Google, Here and Mapbox. BetterPlaces always applies Normalization.
  • Some geocoders are stronger in rural areas, namely Here have really good data there, however Google offers a higher amount of successful geocoding results.