Pages

Tuesday, May 24, 2016

Big data crashes the Tour de France - insight on wheels or plain confusing?

Cycling has turned into a test for computing. Can big data possibly explain the world's most most complex sport?


At the end of the week when big data feeds and handlebar-cams collided with the Tour de France for the first time in history it’s time to ask ‘la question évidente’ - is the world any the wiser?
So far Dimension Data’s detailed feed has attracted the most attention, with puzzled TV presenters and sports journalists divided about the insight offered up by this startling new world.
On the one hand, fans were told just how fast sprinters really go with super-sprinter Andre Greipel hitting 69.44km/h on the intermediate sprint on Stage 3, to pick out only one nugget, the sort of trivial but fascinating data that would up to now have been pure guesswork.
On the same stage on a downhill segment, Lars Boom of Astana reached an extraordinary 108km/h at the 144km mark while on Stage 7 winner Mark Cavendish ambled along at an average speed of 42.74km/h.
All fascinating stuff no doubt but there has been the odd hiccup, joyfully pounced upon by a surprisingly sceptical press corps.
On stage 7, for instance, mid-stage data from GPS transponders located under the riders’ seats appeared to show sprinters Cavendish, Greipel and Peter Sagan were three seconds apart (about 33 metres at average speeds) despite the TV images showing them as almost joined at the hip. That had the commentators chortling at the failings of AI in the sky.
Meanwhile there have been mutterings about who actually owns all of the data being poured onto Dimension Data’s live tracking website, both from teams and journalists. The answer almost certainly is organisers ASO although the latter business empire appear to have bowed to team sensitivity in keeping some competitive data away from prying eyes.
This will become more obvious in mountain stages where teams and their rivals will be able to work out how many watts a particular riders was burning by calculating distance and time with great accuracy. They’ll even have precise wind direction and speed data and altitudes to make it more precise. In cycling that sort of stuff is gold because it suggests form and strength.
In a data pincer movement of sorts, GoPro’s HERO4 cameras mounted on rider’s handlebars have offered up a totally new view of what is like to ride the Tour from inside the action. These have been used experimentally for a couple of years now but this is the first time the outside world has been given this view of a major race such as Le Tour.
Of course, interest in handle-cams tends to hone in on crashes such as the cyclists-as-skittles footage from Stage 5 although the insight it gave into the treacherous sprint finish is just as illuminating.
Eventually,, the rider speed, position and proximity data will be integrated into the camera footage and Mark Cavendish’s recent complaint that people treat sprints as if they are PlayStation events will have come true in ironic fashion.
But it’s clear that cycling will never be quite the same again and professionals will adjust. In two years they’ll scream loudly if a feed goes down.
For decades cycling has been a strange sporting pauper outside the mainstream but it has finally pulled off a coup. They said it was the sport whose inner secrets were almost impossible to see, too chaotic to understand. The odd wrinkle aside, big data says they were wrong.

Image credit: GoPro/ASO

No comments:

Post a Comment