Introduction
I did not intend to write this article. I was just doing a little creative thinking one day, and what emerged was this article. I have a feeling additional insights and articles will spring from this work in the future, so I decided to write this brief study.
Background
I’ve been working with a lot of data lately (over 3B records for one project this year). When you start thinking in terms of tens of billions of miles of driving in a year, interesting and curious thoughts can emerge because of the pure volume of data being collected. It is mind-boggling.
Today, my natural curiosity got the best of me and I decided to visualize the movement of people around the country. I wanted to visualize the common pathways drivers take when moving around the US to better understand the flux (i.e., movement) of people around the country. Previously I have done this with airline data, which of course shows where the major airports exist. In this case, the origins and destinations of people can be anywhere and the results depend upon the road networks.
What If We Rebuilt Our Interstates To Accommodate Modern Travelers?
Today I wondered about our interstate road networks. I asked myself some questions and decided to see what the answers might look like.
Here are the questions: What if we could erase our road networks and replace them with straight roads, to connect us in the ways that we need to be connected? How much time, fuel, and other resources could we save? I wondered what the roads should look like now (in 2017) compared to the interstates that were designed and built decades ago? I wondered how many roads would be needed and where should they be placed?
To begin to answer these questions, I wondered if I could visualize the current flux of drivers and imagine how a new road network could be built to accommodate the drivers that move between our major cities.
To answer this, I created a movie (a one-month time series animation) that essentially captured the actual starting and ending location for drivers that covered between 50 and 500 miles of driving in a day. I did this for a whole a month, using a small portion of my data. What emerged was an interesting visualization that takes a couple of minutes to develop because of the volume of data. The resulting straight line map reminds me of some type of non-uniform neural network.
I experimented with colors, line thicknesses, and other settings while creating this movie. What I realized was that the black line segments told the story more effectively than anything else I tried. Although I am not showing fine-scale detail in this article, some of the insights I have seen by zooming into major metro areas are fascinating.
Figure 1 is the end game of this exercise. It is clear that much of the city to city driving occurs in the eastern half of the country. I wonder how much of the open land in the western US will develop over the next 100 years or so such that those big gaps in roads will no longer exist. It is also clear that traffic in southern California is severe.
Figure 1 – Straight-line paths between origins and destinations in November 2017. If we had straight roads, this is what they should look like. The density of the black lines indicates the volume of travelers between the points of interest.
By plotting the existing US interstate network (Figure 2), I could see the strong correlation between where people are moving and the existing interstates. By panning and zooming around, I could see regions were additional interstates are needed. These are mostly regions where population growth outpaced the capacity of the existing road networks.
Figure 2 – This is the US Interstate Road Network. It is clear that many of the lines shown in Figure 1 use the interstate system. There are regions where more interstate roads need to be placed. For example, consider west Texas and New Mexico. Can you see a lot of travelers moving between locations without interstates to travel on?
Future Work
One of these days I’m going to determine how much extra driving we do every day due to the curvilinear nature of roads. I’m going to compare actual driving distances to the straight line distances that are calculated between starting and ending locations of a trip. This will be a big computational study that will require an Alteryx analysis, but it will help me quantify where we could use more roads. It will also help me understand how inefficient our current roads are for moving us around.
