hacking idea

27 views
Skip to first unread message

Brian DeRocher

unread,
May 22, 2014, 4:58:32 PM5/22/14
to nova-b...@googlegroups.com

Hey there,

 

It was nice seeing some of you at the Bike Hack Night. Thanks for coming out to hear me blabber about cycleways in OpenStreetmap! I'm not 100% sure what happens in the NoVa Brigade group, but i'm learning quick, and i think this idea may be interesting to some of you. At least it would be nice to get some feedback and thanks in advance.

 

It's about gerrymandering and how to use Census data plus a clustering algorithm to build the voting districts.  While working on a k-means clustering activity for something at work, this was published: http://bmander.com/dotmap/index.html


Sorry looks like the Dotmap is busted.  See http://geographyeducation.org/2013/01/01/u-s-census-dotmap/


And it got me thinking, i could apply clustering to these dots, i mean people!  K-means goes like this: http://statistical-research.com/spatial-clustering-with-equal-sizes/


But i'd do it for 1 county at a time. The algorithm uses a "distance" function.  And i'd hope to use a walking distance instead of how the crow flies, type of distance function.  And that's where my instance of OSRM comes into play (http://mappingdc.org/osrm).

 

Thoughts?  I would be glad to do all the coding, but i need help with ideas, some research, and publishing the results.

 

I'd like to attend this hackathon on May 31, but not sure if i'll have time. I may have family obligations to take my daughter to the playground, and of course collect details for City of Alexandria ;)

 

Brian

 

PS: A few years ago DC gov had a small web app to build districts. Does anyone know where that is?

 

--

http://brian.derocher.org

http://mappingdc.org

http://about.me/brian.derocher

Brian DeRocher

unread,
May 22, 2014, 5:01:21 PM5/22/14
to nova-b...@googlegroups.com

Brian DeRocher

unread,
May 31, 2014, 1:27:38 PM5/31/14
to nova-b...@googlegroups.com
I can't attend the hackathon in person today, but i'm working from home. My project is to combine the Census Dotmap with k-means clustering to generate voting districts, in an attempt to eliminate gerrymandering.

My project is located here https://github.com/openbrian/districtbuilder I'll update the README soon.

If you want to help me out, just email me. I'll be checking in sporadically (vis. when the kids are napping).

--
Brian DeRocher
Reply all
Reply to author
Forward
0 new messages