SR 400 After Tolls -and- Traffic Data for the Win!

A couple weeks ago, Jenn texted me at work and asked if I’d heard that the traffic on SR 400 had increased by 30% after the tolls came down. Apparently one of her co-workers had heard that number recently and wanted to know if it was true.

I had not heard, but I have resources and I could find out.

First off, some background for those of you not native to the area. SR 400 is an fully-access-controlled expressway (interstate, without being an interstate) that was built with bonds that were paid off by tolls on the roadway. SR 400 is a critical north/south link to the northern Atlanta suburbs and some of the most prodigious economic centers including Buckhead, Perimeter, and bedroom communities to the north. It is highly congested at and north of the I-285 junction and is pretty damn congested south of there. The map here shows the location of the toll and SR 400’s position in the Atlanta arterial system.

SR 400 Location Map

On November 22, 2013, the last toll was paid and the toll booths were shut off. What’s happened since then? Hard to tell, really. I mean, it’s only been a couple months and we had two major holidays in there. It’s difficult to get a clear picture. However, I’ll state that so far, traffic has increased by about 7% daily and 23% if you look at the correct hour.

But, Bill! Just tell me how much traffic has increased so I can have a news blurb.

It’s not that simple. Here, I’ll show you why. From here on in, this is exceptionally tl;dr1 for people who aren’t nerdy about data. You are warned. We are descending into the depths of data analysis, why traffic data is hard to look at, and squishy numbers and assumptions.

First a graph. This graph tells you what I just told you.2.

SR 400 North of Tolls Graph

What is all this? This is the change in traffic volume at a point north of the tolls but south of I-285 after the tolls came down. This shows how a particular hour of the day, going from midnight on the left side of the graph to the following midnight on the right side of the graph, has changed. The big lines show the magnitude of the change with one being the northbound side of SR 400 and the other being southbound. The bar graph underneath them shows the percent difference of each hour compared to that hour prior to the tolls being shut off. Just by scanning the bar graphs, you can see that the 7% increase is about right and the peak change is about 23% around 5:00 and 7:00 AM. The biggest surprise to me, however, was the significant drop in volume, northbound, during the 5:00, 6:00, and 7:00 PM hours. What the hell? Traffic actually dropped after the tolls came down? Especially during what is colloquially termed “Evening Peak Hour” and also colloquially called “Traffic Hell”?

Yes, apparently it did.

However, and we’re going to sideslip into the lesson of “It’s easy to lie or obfuscate with graphs”, this graph doesn’t actually tell you very much. You could draw all sorts of conclusions from this, but without knowing how I generated it, or comparing it to a similar set of data, I could easily mislead you into a false premise that would have you running around in circles. Lets look at another graph.

SR 400 North of Tolls Graph

This graphic shows the average daily traffic over the whole of 2013 compared to the average daily traffic over 2012 at the same location, by direction.3 I also included that same bar graphic that shows the percent change by hour. Take a moment and look at that graph and see if you can see what I’m seeing.

According to that graph, in 2013, traffic was lower than 2012. Overall, it was 1% lower southbound and 2% lower northbound. Now, notice the 5:00, 6:00, 7:00 PM hours. See how the 2013 northbound traffic shows a significant decrease compared to 2012? What’s up with that?

No idea. Well, I have some ideas, but they’re just thoughts. I don’t have any real, informational basis to answer that question. We’ll come back to that.

Let me now tell you some more about what you were looking at in the first graph. All I said was I was comparing traffic volumes before and after the tolls came down. But what exactly was I comparing? I could have chosen to compare November against December, but that’s a bad idea for reason I’ll note in the footnotes.4 I could compare the average Monday, Tuesday, Wednesday (etc.) from the previous year to my December, Monday, Tuesday, Wednesday, but I decided that was too much work. I could compare each particular day (12/1, 12/2, 12/3) to it’s corresponding day the year before, but that won’t work.5 So, what I did was compare each Monday, Tuesday, Wednesday, Thursday, and Friday in December of 2013 to its corresponding day in December of 20126. Some days had to be thrown out around Christmas. Others don’t exist in the dataset because the quality assurance people at GDOT decided they weren’t good data. Then I did some number crunching with Excel and voila! Graph number 1.

All of these numbers and graphs are interesting and give people something to talk about, but they certainly don’t fill in the little details. For example, I’m comfortable saying that traffic has increased due to the tolls coming down. That was the expected result. However, I’m very interested in why 2013 traffic seems to be lower than 2012. I did not expect that result at all. Since 2007 and the housing bubble bursting, traffic nationwide has gone down and continued that trend, or remained flat, for quite some time, but in the last year or so, it’s started moving back up.

And I’m very very interested in that sharp downturn in PM peak hour traffic that the northbound graph is showing. What’s going on there? Was there construction that reduced the capacity of the roadway?7 Did I-285 traffic turning onto SR 400 northbound increase and cause a backup to the south?

Again, I don’t know. What I do know as of right now is that this was a long term event. I’ll have to go back into the data tables to see where the volumes started to tail off and talk to some of the GDOT folks I know to see if there were some construction or other activity that may have caused it.

This sort of data-crunching is extremely useful for seeing long term trends, and short term effects of decisions or construction. It’s also extremely laborious unless you have the systems set up to do the work for you. It’s also easy to twist the results into meaning what you want it to mean, but if you’re a connoisseur of internet information or politics, you probably already know that.

I had fun answering the question of Jenn’s coworker, and was lucky there was an automated count station in the vicinity that I could look at. I got the data from the GDOT Traffic Polling and Analysis System and if you go through the link you’ll see that there aren’t that many full-time automated traffic counters out there. It’s not like there’s one on every road or intersection.

This sort of thing is one of the reasons I enjoy my job as a Traffic Engineer and I’m happy to have had a project to play with on my free time.

  1. Too Long, Didn’t Read. []
  2. Although the precise values of 7% and 23% I pulled off of the data table, which I have helpfully not supplied here []
  3. It’s important to separate the directions of travel because, in essence, these are two different one-way roads. While there are circumstances when it’s necessary and useful to combine the volumes together, for the purposes of looking at the changes in driver behavior, you should not. []
  4. November and December are tough months for smooth traffic analysis because of the holidays. Nobody is on the road on Thanksgiving day but everybody is on the road the days around it. The shopping season is upon us, and don’t forget there are two regional malls within the “reach” of this part of SR 400. Then Christmas appears. What a mess. []
  5. You need to compare similar days, such as 12/1 this year being a Sunday, but last year it was a Saturday, and don’t forget that this year it was the Sunday right after Thanksgiving. Last year it was a week later. []
  6. See last footnote for the fancy dancing on that comparison. []
  7. If a roadway is at peak capacity, it will show on graphs like these in the peak hour bu not exceeding a certain level and plateauing. Anything that effects the number of vehicles that it is possible to squeeze through the roadway will cause that plateau to rise or fall. []
This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>