This week, Betsy Mason published a great piece with National Geographic about ‘megaregions’ within the U.S. The article discusses recent research conducted by Garrett Nelson of Dartmouth College and Alasdair Rae of the University of Sheffield that uses census data to map commuter paths and ultimately to determine clusters of interconnected cities. These ‘megaregions’ are key as urbanization continues—particularly from policy and planning perspectives.
The interesting thing about this particular research is that they conducted it both on a visual and on an algorithmic level. They started out by taking all of the commuting data and drawing it on a map. They could estimate from there what seemed to be connected—what the hubs appeared to be. Then they went a step further and analyzed the data based on an algorithm tool that was developed at MIT. That tool took the strength of the connections between nodes into account, ignoring the physical locations.
To show the contrast between the two methods, Mason uses the Twin Cities as an example. In the visual analysis, it looks like the Twin Cities are the hub that ‘extends outward concentrically.’ Yet the algorithm analysis zooms out and finds other regional hubs that are connected to the Twin Cities’ hub and prove that the latter is the largest of multiple hubs rather than the single hub of the region.
The visual illustration of the twin cities is on the left and the algorithm-based analysis is on the right. Illustration by Garrett Dash Nelson and Alasair Rae. Plos One.
Yet the algorithm also ended up with some bizarre results—the lack of geographic consideration resulted in bizarrely large splotches (hubs) and delineations (such as between Connecticut and New York) that didn’t take cultural continuity into account.
Thus the researchers combined the two methods: they started with the algorithm and then corrected based on the visual results. The final product can be seen below.
This research is valuable both in what it tells us about megaregions, but also as a lesson for how to approach planning. Both the visual and the algorithmic approaches had weaknesses and the most accurate result was a combination of the two. The same goes for city and notably transportation planning. The best result always comes from a combination of visual analysis and technological approaches.
What do you think about the regions below?
A combination of visual and mathematical approaches. Illustration by Garrett Dash Nelson and Alasdair Rae. Plos One.