Canadian American data adventure stats

Quick look at the vehicle passing data from my 3,757 mile bike ride over 3 weeks from Edmonton, Canada to Hoover, Alabama through 2 Canadian provinces, 12 US states, and 115 counties.

15,000 vehicles

That’s about how many vehicles passed me on this adventure while I had my radar enabled. This equates to just over 4 vehicles passing me every mile or 1 vehicle every minute. (4.06 vehicles/mi and 0.95 vehicles/min). But the overall average doesn’t reflect the significant difference between Canada and the USA. Here’s a breakdown:

Why was Canada busier than the USA (5.9 vehicles per mile vs 3.6 vehicles per mile)? The reason was route choice. I spent a lot of time on Trans Canada Highways 16 and Highway 1, easily the two busiest roads in Western Canada, because I had mechanical problems fairly early that kept me from wandering onto the gravel and dirt roads that were plentiful in the area. This meant I spent a lot of time on the busier paved roads. See all the pictures and read all the details on my cycling blog post: https://toonecycling.com

But what about speeds? How did Canadian drivers compare to American drivers? The table below give the results. Note that I needed several tables. The first is the overall speeds, but that is not a perfect comparison because I was on busy Canadian highways for almost of my route through Canada. In the USA, I had a wider mix of rural roads and major highways (interstates and busy US highways). The second table only includes interstates and major highways. And the third table excludes interstates and major highways. All three tables are sorted by higher average vehicle speed.

Table 1. All roads.

Table 2. Interstates

Table 3. Non-Interstates.

SPEED ANALYSIS

Tables 1, 2, and 3 compare average and relative speeds of vehicles on all types of roads, Interstates only, and Non-Interstates only. The results seem to indicate that Canadian drivers drive faster than American drivers. Yes, they do indeed, for the roads I routed myself on to stay on pavement - again nearly all Trans Canada (Interstate) or State highways - very few secondary roads at all. Only 130 passes out of nearly 4500 occurred on roads classified as secondary.

Looking at Interstates only (Table 2), Americans are back in the fast lane. Speed limits could factor into this. The only Interstates I ended up on were I-90 and I-25, both with speed limits of 80mph for most of the sections I was on. Some of the more mountainous sections had 70mph or even 65mph, but those sections were short. In Canada, most of the Trans Canada highway sections I rode, had speed limit of 110kph, which I believe equates to just under 70mph.

(LEFT) Alabama State Highway 14 is classified as a PRIMARY road. Note the small shoulder with a rumble strip preventing cyclists from even using what little is there. (RIGHT) Alberta state highway 93, also known as the Icefields Parkway, is classified in OSM as the same kind of road as the Alabama highway.

Table 3 (Non-Interstates) is interesting. A state highway in Canada tends to have a very wide shoulder. The Icefields Parkway (above, right) is a good example. Tourists and bus drivers drove extremely fast on this road. I never felt in danger because of the wide shoulder. On roads like Alabama Highway 14 (above, left), speeds may have been a little bit slower, but they feel much more dangerous because there is no room for error. The average passing speeds over the long 125 mile Canadian Icefields Parkway was exactly 50mph (80kph) with the speed limit varying from 50kph to 100kph. I was passed by 930 cars (and tour busses) but no commercial trucks as they were not allowed on the highway. This explains why the Canadian non-interstate speeds are significantly higher than the USA non-interstate speeds, which groups a wide variety of smaller roads many of which have no shoulder at all once you make it to the Southeastern portion of the USA. I do have both the ability and desire to break the USA up into segments and re-run the analysis, but unfortunately I do not have the time right now! I may save that for a future post that uses not just my own data, but data from all the riders using my app from both Canada (435 riders) and the USA (2,123 riders).

A standard deviation anomaly

Table 4 below shows the standard deviations for all three values (absolute speed, relative speed (i.e., speed differential), and rider speed) for overall and the various road types. I am wondering if there is something interesting there. The deviation in my own “rider speed” is significantly higher in Canada, but the standard deviations between absolute passing speeds and relative passing speeds are identical! Contrast this with the USA standard deviations, where my rider speed standard deviation is lower and the standard deviation difference between absolute and relative speeds is higher!

Table 4. Standard deviations. Why are the Canadian absolute and relative speed standard deviations for vehicles the same when it is a calculated value that takes into account rider speed? What does this say about driver or rider behavior?

My increased rider speed standard deviation in Canada definitely stems from there being more climbing per mile in Canada (45 ft/mile) vs USA (33 ft/mile). But how does that equate to an equal standard deviation between the absolute speed and relative speed in Canada? Shouldn’t there also be a difference in absolute vs relative speed since that calculation takes into account my rider speed. I guess I don’t have a strong enough grasp of standard deviation to understand the implication. But my “hunch” is that this somehow shows a difference in Canadian driver behavior vs USA driver behavior. Or again it might point back to the uniformity of roads I was riding in Canada (i.e., fast and busy). Still, I think there is something interesting in these standard deviations. If you have an idea and explanation, please give it to us in the comments!

Finally, what about lateral passing distances? Well, I hate to disappoint but you will have to wait until my next post. I am working hard to analyze nearly 200GB of lateral passing data, which includes photos of cars as they pass, LiDAR depth maps of the cars, and metadata such as phone orientation. I am currently processing this data to refine the LiDAR distances to what I believe to be the most accurate possible and will post that soon. In the meantime, here is a teaser of what I’m working with!

Passing car being detected at 0.910m lateral distance at its closest point.