Wireless and Digital Services
Results are in from the 2017 edition of the Boston Marathon, and it looks like Boston’s old tradition of chasing women down the course and not quite catching them is alive and well…
I’m an algorithm developer in our algorithms and analytics group – and I like to run marathons. We apply algorithms, machine learning and artificial intelligence to turn data into useful insight and to solve the toughest problems for our clients. But it’s not all serious work: I thought it would be interesting to dig in to the Boston Marathon data and see if some of the marathon clichés are true.
(If you’re a Boston techy looking for a new challenge, take a look at our current vacancies.)
We harvested GPS data for about 10,000 runners who ran the race from 2014 to 2017 and had a look at how they paced the race. It’s widely held that women pace themselves more evenly over long distances than men (here’s some research on elite runners and here’s some research on the rest of us). Once you get past the judgemental tone of some of this work – heavily in favor of conservative, comfortable marathon running – you wonder, is it true that women pace more evenly? More particularly, is it true for Boston, where there are special qualification criteria and no pacers? And if so, how far into the race does the difference emerge?
We can guess how fast a person planned to run a marathon by looking at how fast they ran the first half, then assuming their average pace over the first half was their target pace. This approach may be a little bit out for Boston, given the elevation profile, but it’s near enough for a quick comparison of sub-populations of runners.
For each runner we worked out their speed as a percentage of their assumed target speed. The lines on the plots below are deciles of this relative speed, from the tenth centile up to the ninetieth, as a function of distance along the course. At any given point on the course, exactly 80% of runners are in the shaded band.
At half way, most runners are within 5% of their target speed. Then they start to get tired (and there are hills to go up):
By mile 23, more than half the runners have dropped at least 10% below their target pace. The course is mostly downhill from mile 21, so no excuses there: these guys are having a bad time.
If you make the same plot again, but separate runners by gender, you get this:
The blue band at the bottom right (pic above) is a large group of men who slow down catastrophically in a way that not many women do – a divide emerging gradually throughout the second half. A suspicion that you’d see the same difference between older and younger runners appears to be unfounded: There’s a minimal difference in speed drop-off between the under 30s and over 50s:
So when you’re running the Boston marathon, wondering who to latch on to as your unofficial pacer, pick a woman, and don’t worry how old she is. If you’re a spectator? Find your way to mile 23 if you want to see the battle between mind and body in full effect…
We’re recruiting for a huge variety of exciting roles at our office in Boston, MA. Check out the full range here.