Is this the structure of Chicago?

    Prev Next

    It's not the Chicago of my experience, but almost all of the Chicago of my experience was just that tiny spot at the bottom right.

    The curve heading to the southwest seems especially suspect for not corresponding to any particular road or transit line.

    Data from the Twitter streaming API (10000 points, 30000 vectors). Base map from OpenStreetMap, CC-BY-SA

    A set of the same technique applied to other cities

    karlsbad, SergeyKD, ian_dewbecker, and 31 other people added this photo to their favorites.

    1. Oculator 27 months ago | reply

      It would be interesting to see an overlay of this on your ethnicity map. Seems like folks don't Tweet much driving down the Dan Ryan, especially as compared to the Kennedy and Edens. Thanks for putting up all this beautiful work.

    2. Eric Fischer 27 months ago | reply

      Keep in mind this is trying to be a map of travel from locations to other locations, not of individual tweets. The individual tweets are just to guide the paths. I think what you are seeing here is mostly a lot of travel between O'Hare and the Loop, not a particular tendency to tweet while driving on that route. (Also, the Edens is hardly represented here at all. Those two big routes to the north and northwest are Clark Street and Milwaukee Avenue.)

    3. biokinetic 27 months ago | reply

      also note that the cta blue line runs down the middle of the kennedy from o'hare to the loop. the southwest leg you're seeing is the elevated orange line that runs from the loop to midway. i'd bet that most people are just tweeting on the train to or from a flight.

    4. find myself a city (1001 Afternoons in Chicago) 27 months ago | reply

      Another product of the University of Chicago? (That is, did you graduate from U of C?)

    5. shadowsinspace 24 months ago | reply

      Eric - I think you might enjoy this upcoming show I'm in, for real not just plugging cause I'm in it... :)
      Natural Networks: Systems of scale, thought & unseen connectivity. Venture into the borderlands of the abstract on.fb.me/Iz3X6u

    6. fhughes88 18 months ago | reply

      This map says only that higher SES categories tweet while using transit. It says nothing of demand for transportation.

    7. Eric Fischer 18 months ago | reply

      If you look closely, the routes are mostly actually assigned here to major commercial streets, not to the transit lines themselves. Tweeting from transit is a relatively minor component.

      You are probably right about low economic status being underrepresented, but I don't think you are right about social status. If you read the Twitter public stream for a while, you can see that US tweeting is dominated by young speakers of stigmatized dialects. You do point out the need to study this bias systematically though.

    8. metaroll 10 months ago | reply

      Hi Eric, did you apply some path aggregation or bundling algorithm to the data? E.g. something like Floyd–Warshall?

    9. Eric Fischer 10 months ago | reply

      The path aggregation is by Dijkstra shortest path routing, but I think the Floyd-Warshall result would be equivalent in this case.

    10. metaroll 10 months ago | reply

      Thanks! Did you calculate all shortest paths between the 10000 points or did you just use a subset?

    11. Eric Fischer 9 months ago | reply

      I calculated the shortest paths between all 30,000 origin-destination pairs. I'm not sure if it actually ended up calculating the complete network in the process.

    keyboard shortcuts: previous photo next photo L view in light box F favorite < scroll film strip left > scroll film strip right ? show all shortcuts