Yesterday was so fun, but hectic! I gave a talk on http://redistrict.science [preprint: https://arxiv.org/abs/1711.04640 ] and it was also coincidentally #GISday. I was focussing on how I did what I did using Javascript, Cython, C, Python, GIS, etc. Today I'd like to talk about the science.
-
-
The redder a state is the more our method improvesd it. As you can see from the map (and based on analyses we did), bigger/more populous states are redder. Interactive maps can be found at: http://redistrict.science and the preprint: https://arxiv.org/abs/1711.04640 has many more details.pic.twitter.com/AeESSkcidx
Show this thread -
Take top row [taken from the preprint https://arxiv.org/abs/1711.04640 ], that's Iowa (a), our algorithm (b) improves on it, even tho they are trying their best. That's the thing! Even when it's done by impartial committees, computers still district better! More: http://redistrict.science pic.twitter.com/1oYxiAMbMK
Show this thread -
In bottom row, you see the very sorry state (pun intended
) North Carolina is in! Charlotte (c) is cracked so much, reaching so far into the countryside. our solution shows how 2 urban districts for Charlotte make sense.
More: http://redistrict.science http://arxiv.org/abs/1711.04640 pic.twitter.com/tTqSxbRhCJ
Show this thread -
Another
in Utah. West Valley & Salt Lake Cities (a) are cracked so hard the district goes all the way to other end of state! Same with Provo! Our solution shows how you can fairly give the cities districts of their own.
More: http://redistrict.science http://arxiv.org/abs/1711.04640 pic.twitter.com/FEIqgpQirx
Show this thread -
A really fun thing happens in Arizona (d)! it's horrible in (c) but in our solution you get a
district because our algorithm really wants to give Tuscon its own district! In real world it's cracked 3 ways! Three!
More: http://redistrict.science http://arxiv.org/abs/1711.04640 pic.twitter.com/wl2n2ZPUGP
Show this thread -
Finally, some states really are good.
Omaha, Nebraska has it's own district (e) and then increasingly suburban areas have theires, and then rural theirs. Which is THE SAME as our algo (f) did just more smoothly.
More: http://redistrict.science http://arxiv.org/abs/1711.04640 pic.twitter.com/5UNEsay5KC
Show this thread -
Now I'd like to talk about the algo: I ran our version of k-means on Census data to http://redistrict.science each state.
@DrLoveBC came up with a type of k-means that weights each cluster/district by its population. https://arxiv.org/abs/1711.04640#MachineLearning#ArtificialIntelligencepic.twitter.com/lqk6DxqE4X
Show this thread
End of conversation
New conversation -
-
-
Thank you so much for helping on this. We need all the tools to untangle this mess.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.