GeoCoding our Customers

My experience with Tableau is that you can usually do some basic Spatial stuff, but you have to have zip codes or lat/long pairs already in the data source. One cool thing with Tableau is it has the ability to pull in census data as an overlay. (Or at least in the version I was working in a year or so ago)  I would view Tableau more as the reporting/visualizing piece, not the actual geocoding piece. Might be possible with R, but I haven't tried it. Wonder if anyone else has had a better experience?

I have used ESRI's ArcGIS for geocoding and routing, network analysis, etc. it's amazingly powerful, and looks like they have non profit licenses available for free. It is definitely the cadillac of options out there. They also have some JavaScript and python libraries available that are nice and allow for web and standalone scripting.  I also like their business analyst online tool, but haven't used it specifically for geocoding.

http://www.esri.com/nonprofit/whats-included

an open source option that is on my list to check out is q-gis (qgis.org). Its got a less friendly interface and a somewhat hefty learning curve, but I have heard some great things about it. 

There are also spatial modules available for Postgres and hive/Hadoop databases that are pretty powerful, and easy to use. I have not tried working with Microsoft's spatial options however. Would love to know what other's experiences are with it!

https://github.com/Esri/gis-tools-for-hadoop

http://postgis.net/

https://docs.microsoft.com/en-us/sql/relational-databases/spatial/spatial-data-sql-server

Melanie,

Thanks so much for your reply.  Some really great resources there. 

• I also have run into some of these geocoding features of Tableau. 
• The non-profit licenses from ESRI's ArcGIS sound very interesting.  Has anyone taken them up on this offer? If so what have you found?
• I have heard about the Geolocation components of Postgress SQL, has anyone had an opportunity to try this.  

Since my 2015 post, I have done some more work on Geocoding.

As the first challange, one needs to geocode records.  There are a number of ways to do this.

• One can geocode by postal code. 
  • This can be OK, however, you do have to be careful about the data sets the method is using.  The free ones seem to be less accurate, I've run into some sets that will use very few locations for a number of ZipCodes in the US.  Particularly in big cities.  For example, all of Brooklyn can show up as a single or very few location in some of the data sets.  
  • Apparently, in the UK you can buy access to postal code Geo location data.   However, it is apparently very expensive and produces very small geographic areas due to the geographic coverage of each postal code in the UK.
• Microsoft Power BI has some geolocation capability.  This is similar to Tableau.
• The Knime platform has a number of nodes for geolocation including
  • the Palladian Geo Nodes these use freegeoip.net webservice. Google API, MapQuest, Mapzen
  • There is also a node for geocoding with Open Street Maps
  • There is also geolocation of IP addresses which can be very cool.
• In R I have run into the following libraries:
  • ggmap 
    • this uses google map API as the back end.  However, you are limited to 2500 records a day.   And in some cases, you have to sign an agreement that you will not store this data.  But the location data produced is very good.
  • RgoogleMaps
    • Similar to ggmap
  • zipcode
    • This is a postal code based geocoding method that users the CivicSpace Database (August 2004) and augmented by Daniel Coven's federalgovernmentzipcodes.us web site (updated January 22, 2012)
  • There are others...

What projects are folks working on where geolocation would be helpful?