Figure out a sustainable way to automatically geocode addresses

[fureigh]

...so #4 can move forward.

Current CartoDB plan has geocoding credits cap.

[daguar]

The Texas A&M geocoder works fairly well: http://geoservices.tamu.edu/Services/Geocode/

Another option would be to use SmartyStreets ( http://smartystreets.com/ ), which gives free credits to non-profits and I know has given some already to OpenOakland. cc @spjika

On Wed, Oct 15, 2014 at 11:01 PM, Fureigh notifications@github.com wrote:

...so #4 https://github.com/fureigh/ebt-near-me/issues/4 can move forward.

Current CartoDB plan has geocoding credits cap.

— Reply to this email directly or view it on GitHub https://github.com/fureigh/ebt-near-me/issues/46.

Dave Guarino Consultant, Health Vertical (2013 Fellowship alumnus) Code for America http://www.codeforamerica.org/ dave@codeforamerica.org LinkedIn http://www.linkedin.com/in/daveguarino/ | GitHub https://github.com/daguar/

[spjika]

Smart streets turned out to be super inaccurate, we ditched them On Oct 16, 2014 6:41 PM, "Dave Guarino" notifications@github.com wrote:

The Texas A&M geocoder works fairly well: http://geoservices.tamu.edu/Services/Geocode/

Another option would be to use SmartyStreets ( http://smartystreets.com/ ), which gives free credits to non-profits and I know has given some already to OpenOakland. cc @spjika

On Wed, Oct 15, 2014 at 11:01 PM, Fureigh notifications@github.com wrote:

...so #4 https://github.com/fureigh/ebt-near-me/issues/4 can move forward.

Current CartoDB plan has geocoding credits cap.

— Reply to this email directly or view it on GitHub https://github.com/fureigh/ebt-near-me/issues/46.

Dave Guarino Consultant, Health Vertical (2013 Fellowship alumnus) Code for America http://www.codeforamerica.org/ dave@codeforamerica.org LinkedIn http://www.linkedin.com/in/daveguarino/ | GitHub https://github.com/daguar/

— Reply to this email directly or view it on GitHub https://github.com/fureigh/ebt-near-me/issues/46#issuecomment-59455813.

[daguar]

Oh shit, good to know. Thanks Spike.

On Thu, Oct 16, 2014 at 7:02 PM, Spike notifications@github.com wrote:

Smart streets turned out to be super inaccurate, we ditched them On Oct 16, 2014 6:41 PM, "Dave Guarino" notifications@github.com wrote:

The Texas A&M geocoder works fairly well: http://geoservices.tamu.edu/Services/Geocode/

Another option would be to use SmartyStreets ( http://smartystreets.com/ ), which gives free credits to non-profits and I know has given some already to OpenOakland. cc @spjika

On Wed, Oct 15, 2014 at 11:01 PM, Fureigh notifications@github.com wrote:

...so #4 https://github.com/fureigh/ebt-near-me/issues/4 can move forward.

Current CartoDB plan has geocoding credits cap.

— Reply to this email directly or view it on GitHub https://github.com/fureigh/ebt-near-me/issues/46.

Dave Guarino Consultant, Health Vertical (2013 Fellowship alumnus) Code for America http://www.codeforamerica.org/ dave@codeforamerica.org LinkedIn http://www.linkedin.com/in/daveguarino/ | GitHub https://github.com/daguar/

— Reply to this email directly or view it on GitHub https://github.com/fureigh/ebt-near-me/issues/46#issuecomment-59455813.

— Reply to this email directly or view it on GitHub https://github.com/fureigh/ebt-near-me/issues/46#issuecomment-59457000.

Dave Guarino Consultant, Health Vertical (2013 Fellowship alumnus) Code for America http://www.codeforamerica.org/ dave@codeforamerica.org LinkedIn http://www.linkedin.com/in/daveguarino/ | GitHub https://github.com/daguar/

[lippytak]

Some pros for Google Maps for y'all to consider:

• Nice API (duhh)
• Geocoding API supports 2500 requests/day and 5/second out of the box for free.
• We could get 100,000 requests/day and 10/second with "Google Maps API for Work" < which can be discounted for nonprofits
• And the FAQ suggests we can get infinite for free:

Note: Applications deemed in the public interest (as determined by Google at its discretion) are not subject to these usage limits. For example, a disaster relief map is not subject to the usage limits even if it has been developed and/or is hosted by a commercial entity. In addition, qualifying nonprofit organizations who demonstrate a need for increased geocoding limits or internal usage of the Google Maps API may apply for a Google Maps API for Work license through the Google Earth Outreach grants program.

[fureigh]

Nice! Thanks, all! The Texas A&M link above 404ed, but with HTTPS it works, and looks like a solid option to pursue: https://geoservices.tamu.edu/Services/Geocode/

(More options listed there, should they turn out to be necessary: https://geoservices.tamu.edu/Services/Geocode/OtherGeocoders/)

@lippytak, the Google Maps API terms of service prohibit using Google's geocoder for "any purpose other than obtaining locations that will be displayed using the Google Maps APIs."

...locations obtained using the Geocoding Web Service may not be used by any other application, distributed by other means, or resold.

It'd certainly be possible to rework EBT Near Me to use Google Maps instead, but if possible — and it should be possible — I'd rather not have that switch be motivated by the geocoding need.

[lippytak]

Ahh, lame! Good point. The search continues!

[lippytak]

It's making me feel dumb that I can't find limit or pricing info but someone should look into https://www.mapbox.com/developers/api/geocoding/

[lippytak]

And a bit more info on CartoDB - https://cartodb.com/terms#map_views:

So if we upgraded our plan we could geocode the ~50k ATM addresses for about $350 (15k free + $10/1000 additional).

@migurski think this is worth pursuing with CartoDB? Do we have a good relationship with them? See above for use case. There's about ~50k ATM locations in CA we're trying to geocode to start.

[lippytak]

I used the [Texas A&M geocoding service](https://geoservices.tamu.edu/ for the first 100 and it doesn't look promising:

[spjika]

Point new to the data file, like to see what it does in our internal system On Oct 19, 2014 3:27 PM, "Jake Solomon" notifications@github.com wrote:

I used the [Texas A&M geocoding service](https://geoservices.tamu.edu/ for the first 100 and it doesn't look promising: [image: screen shot 2014-10-19 at 3 25 14 pm] https://cloud.githubusercontent.com/assets/2533112/4694980/0c554154-57df-11e4-94f9-0fdbae4a4e6d.png

— Reply to this email directly or view it on GitHub https://github.com/fureigh/ebt-near-me/issues/46#issuecomment-59668422.

[lippytak]

Full (~50k): https://www.dropbox.com/s/o91pmugdx73lvmj/20140908_cash_access_monthly.csv?dl=1

First 100: https://www.dropbox.com/s/k8pd4iss8yt3j9e/atm_first100.csv?dl=1

[migurski]

I don’t see much indication from Mapbox of where their data comes from.

Google is likely to be the best option available, you’d just need to switch to their client side API. Any reason not to do this?

[lippytak]

Thx @migurski!

Looks like it may be the best option because accurate location is the primary value here. I think we just didn't want to be constrained to using the google maps front end with all the other map sexiness these days.

Would love @fureigh / @alanjosephwilliams to weigh in when in on the pros/cons of switching to google maps.

[migurski]

The other map sexiness is mostly interesting for reasons I think are unrelated to this project: marginal customizability, extensible interactions, and tighter aesthetic control. For accurate dots on accurate maps, there is nothing sexier than plain old Google.

[lippytak]

@fureigh hey hey - so we've gotten pretty far on the gmaps fork and so far we haven't stumbled upon any blockers or compelling reasons to change back to CartoDB (or something else). There are a couple outstanding features before I'd be comfortable deploying but I wanted to ping you to see if you had other thoughts on the carto/gmaps tradeoff. Thoughts? Are we missing anything?