Home Resume Links Blog


Visualizing Triathlon Performance

September 25, 2009

I ran my first triathlon about a month ago. I'm not sure what a normal approach might be for other people doing the same but, being the data geek I am, I ended up with handwritten calendar-based training plans and grids of statistics that my bike computer spat out. I even started the whole mess with running a basic regression analysis on triathlon times I found on the web. That was enough to start training but surely there's more of a story to be told.

It all started this spring when a friend of mine organized her annual duathlon. I've skipped out of a handful of these before because I never got a bicycle after I left my Huffy behind in junior high. I had no good reason to participate in the du this year but I was out of excuses not to since, just last summer, I bought a bike with road tires.

I was assured that if I could run ten-minute miles I wouldn't be the last person to finish, so I made sure I could run one mile at that pace and signed up. Then I was afraid that my biking pace would be so slow that I would, in fact, finish last. It seemed reasonable that there would be some correlation between biking pace and running pace (better athletes should generally be better at both events) so I did a regression analysis. I found some data at triresults.com (it's since been locked in a for-subscribers-only section), tossed it in OpenOffice.org Calc, graphed three subsets of the data, and added the built-in regression curve:

It turned out this wasn't terribly useful except to point out that those who run ten-minute miles in an Ironman (a full marathon-worth of running) bike at far better speeds than I'm capable of. I suspect some of this is because I don't have a proper road bike (no clipless pedals or road handlebars, for example) but, really, comparing a novice to an Ironman athlete is plain silly. I ended up learning to keep my biking heart rate in a decent range and called it good. I finished the du, but it was far from graceful.

Skipping ahead to the tri, I finished somewhere in the middle of everybody, in the middle of all men, and in the middle of all men in my age bracket. My run pace vs. bike pace ended up quite close to the regression line in all cases too (the red points are me):

I did better in all three events than I ever did during training; average never felt so good.

I didn't really practice my transitions between events. Looking back on the data, I actually lost a spot in my bracket because of my slow transition times:

With a triathlon under my belt, plus being an Eagle Scout (really, I am), perhaps I'm qualified to work for 37signals or Fog Creek now.