Visualizing SDOT Open Data Bicycle Counts

Presenting an app for Seattle Open Data bicycle counters. Click the link to go to an interactive data tool for bicycle and pedestrian counts using Seattle Open Data from

If I knew how the Internet worked I would embed the app in the Stats on the Street website, but, given my limited knowledge, the existence of this tool is mostly thanks to the brilliance of Hadley Wickham and the folks at Rstudio. The GUI is hosted through Rstudio Shiny Apps, and I pretty much bought the cheapest plan possible, so hopefully some people get some good use out of this, but not too many people or too much use…

Note that SDOT maintains data dashboards for the Fremont Bridge and Spokane St bike counters, but I was curious about all of the available bike counter data. Fans of the show probably noticed that my project to visit the counters individually and write about it got sidetracked last winter (along with basically all of my Stats on the Street projects), but I am hoping to continue this summer. Find here detailed write ups of the Burke Gilman at Sandpoint and the Chief Sealth Trail in South Seattle. While the SDOT dashboards appear to have access to data uploaded daily, the public urls (refer to the “Notes” tab on the data tool) see only monthly uploads. 

I think when viewing all the counters at once there are primarily two interesting questions: how many people travel through each corridor, and how is that changing over time. You can view long term patterns by crossing type in the default “Daily” tab, where selecting multiple crossing types (ex: “Bike North” and “Bike South”) will show the different values in different boxes. The “Raw Data” tab shows counts by hour of day with different crossing types separated by color, and is probably most useful to understand patterns within days. For example, the Spokane St bridge from West Seattle is largely used by commuters, whereas a corridor such as Broadway on Capitol Hill sees bike trips much more evenly distributed throughout the day (as expected, people aren’t taking their beach cruisers all the way past Alki and over the bridge to tangle with truck traffic on 1st Ave S). Finally, the long term patterns can be visualized on the “Resids” tab, which shows daily crossing counts adjusted for weather and time of year. This hopefully helps disentangle the difference between actual changes in ridership and say a nice weather boost. (I think it’s fairly well established that people are much more likely to ride a bicycle when it is sunny and warm, rather than some combination of cold, rainy, and dark.)

With respect to total bicycle traffic, it is unsurprising that the Fremont Bridge wins the title as the bottom of the funnel for bicycle commuters from northwest Seattle to downtown. Other high volume corridors include the Elliot Bay trail, Burke Gilman at Sandpoint, Mountains to Sound I-90 trail, Spokane St Viaduct, and 2nd Avenue cycle track. The neighborhood greenway counters and the Chief Sealth Trail (in the awkward location where the counter was installed) see the lowest volume. This makes sense, though, that in general highest traffic occurs at either chokepoints where a huge geographical area is funneled through a single corridor (Fremont Bridge collects NW Seattle and Spokane St Viaduct collects West Seattle), or routes where people on bicycles enjoy fully separate infrastructure (bike paths, to some extent the 2nd Avenue PBL or Broadway cycle track). I certainly hope, though, that nobody sees the measured counts as some sort of indictment of the Neighborhood Greenways, since they fill a valuable role that has nothing to do with the spandex crowd or funneling commuters from a whole quadrant of the city.

Cribbing from the “Resids” tab, it’s interesting that many of these counters show no obvious time trend. I would guess that, as car traffic congestion worsens during the city’s current economic boom, more people switch travel mode. The most recent census data bear out this hypothesis, as explained by Brock Howell on the Seattle Bike Blog. It would seem as though people are pouring into Seattle more for economic opportunity than convenient motoring, or at least arrive somewhat agnostic to transportation mode, more likely to choose based on cost and reliability than prior notions of what transportation should look like.

(Also, some editorializing: I either walk or ride a bicycle past the Mercer Mess most days and find it remarkable that people continue trying to drive alone down Mercer Street day after day after day. Like, I did it once five years ago and thought, well I’m never doing that again, which I guess means that the giant overhaul devoted tons of resources to more SOV lanes and it’s still a disaster. Really makes you think. It also occurred to me recently at the intersection of 5th N and Mercer that I probably don’t have enough arm to throw a football to someone waiting at the opposite corner, which seems a terrible indictment of urban street design — like this is not a huge suburban intersection or highway interchange but an area by the Seattle Center and Gates Foundation bustling with people on foot.)

Anyway, back to time trends: the Fremont Bridge counter shows a consistent upward trend, but many others are either flat or ambiguous due to data oddities. In fact, a closer inspection of the Fremont Bridge reveals what might be problems with the west sidewalk counter starting last fall. I recall some amount of closures for maintenance (painting?), but I don’t think traffic diversions would quite explain the pattern. The west sidewalk spikes in the resids (crossing counts loosely adjusted for season and weather) seem a bit curious, but I’m not sure whether that was caused by the counter freaking out or some real occurrence of unusual behavior by the people riding bicycles across the bridge. The fact that the east sidewalk counter looks smooth and unremarkable is suggestive of a data problem…

With respect to the other counters showing weak or no time trend, it may just be that a handful of snapshots doesn’t fully capture something as complicated as the aggregate behavior of tens of thousands of people citywide. Anecdotally I feel like I see more people out and about riding bicycles, and even if the overall rate of bicycle ridership remained the same you would expect to see an increase in bicycle traffic as more people pour into Seattle. I think the bicycle advocates/safe street/urbanist people would like to see explosive growth in these numbers, and The Seattle Times and KIRO Radio would like to see numbers so small that they can thump their chests and yell WAR ON CARS every time SDOT breaks out the green paint, but as with most things the actual answer is disappointingly in the middle. Riding bicycles is a real thing that people in Seattle do for transportation and fun, but the SDOT counters do not reveal rapid & enthusiastic adoption, rather a slow burn partially obfuscated by data oddities. (A recent STB podcast had the best description of The Seattle Times that I’ve ever heard, which was something like “good reporting with terrible headlines followed by absolutely crazy editorials.”)

Those are my main observations after a bit of exploration of the data, and I’m sure that there are a ton of interesting nuggets to be found. Here are a few more observations:

The pedestrian counters often seem to have sporadic days of HUGE counts, which I’m guessing represents races/walks/events?The “Remove Values Above” input was basically added to take out the few days of mega counts in this situation to better see the rest of the pattern.

While most counters in the raw data view show a morning and evening weekday commute spike, it’s pretty cool that the Elliot Bay pedestrian counter shows a very distinct lunchtime peak. Much more relaxing to take lunch by the sculpture park and the bay.

The direction on the Fremont Bridge is ambiguous because the counters record the side of the bridge on which the crossing occurred, but not the direction. People bicycle both directions on both sides of the bridge, especially when heading to/from Westlake or the South Ship Canal trail.

It looks like the Oregon St Greenway in West Seattle had data problems that were fixed around February 2015. Same for the counter at NW 58th and 22nd Ave NW in Ballard. The regression model currently fits to all data at once, so the resids (and also the lines on the daily plot) aren’t to be trusted when just viewing after the data fix.

The measured traffic on the Broadway cycle track was way higher in summer 2014 than summer 2015. Not sure what that’s about… I don’t spend enough time on Capitol Hill to have any anecdotal observations or suppositions.

The Mountains to Sound trail at the I90 bridge and the Burke Gilman near Sandpoint apparently see slightly more weekend recreation traffic than weekday commute traffic. In fairness, if you are riding downtown from north of Magnuson Park or east of Lake Washington/Mercer Island that would make for a pretty serious commute.

The greenway counter at 39th St NE and NE 62nd St showed way higher counts in the winter of 2014/2015 than the winter of 2015/2016, even adjusted for differences in weather. I don’t know what that’s about.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s