IPD/Byron counts

A couple of weeks ago, I undertook to try crowdsourcing a traffic analysis of the interesction of Island Park and Byron, working with Kevin O'Donnell on some new tools. I'm pleased enough with the results to try it again, and now Kevin is setting up film I took of the Harmer/Wellington intersection as a second attempt.

So, what did we learn from the first effort? I've attached a spreadsheet to this post below if you want to dig in to the data. Unfortunately, a glitch during the first capture phase meant we had to re-open the analysis engine for subsequent rounds, but we nonetheless received lots of useful data once everything was working perfectly.

• Analysis was captured from 63 IP addresses, and each address analyzed an average of 9.3 clips;
• Each of the 100 clips was viewed an average of roughly 6 times, allowing us to cross-check the results
• Using the average of each clip's views, we counted 77 pedestrians, 68 bikes, and 2311 cars crossing the intersection over the 75 minutes of filming
• Nearly every phase saw drivers run the red light
• We saw a couple of "Pittsburgh lefts" - cars that make a left turn immediately when the light turns green, including a City truck
• There were multiple incidences of cars mounting the curb to go around left-turning cars
• There were only a couple of incidents of crosswalk-blocking

I make no representation that these results are typical of what we'd find over multiple filmings, and in fact I'll go back a few times to get a representative sample. Anecdotally, the Friday morning I shot this video was pretty light traffic.

The attached spreadsheet contains the original data file, a scrubbed version (two results were clearly not accurate), and some pivot tables with further analysis. The "observations" tab is particularly interesting reading, with lots of bad behaviour by drivers, a few cyclists, and even one jaywalking city councillor. In that tab, I've hyperlinked the video numbers to the raw footage if anyone wants to go back to watch.

I am very grateful to everyone who participated. I consider that this was successful enough that both Kevin and I are keen to try it again, tweaking as we go. I'll be sharing the results with the City to get their feedback, as well. In fact, I'm hoping to start a conversation about making the intersection safer based on this data round alone. Thanks everyone.

(The table provided by Kevin is the first tab - any errors of analysis or methodology in subsequent tabs are mine and mine alone. If you see problems, drop me a line so that I can adjust.)