We’re very excited to report that we have been able to add an exciting data set to our repo for the challenge.
Every athlete was wearing a transponder, and we have their positions, acceleration and speed sampled every ten seconds for the majority of the races!
As an example, here’s the file for the Mens 800 final - if you scroll to the bottom, you can see who actually ran how many metres, and how far they were from the rail, at all times in the race…
This should open the door to many novel explorations and visualisations!
3 Likes
@andyrobinson that’s very interesting!
Were the transponders worn on arms or feet? I’m looking at the data of the first athlete and wondering why the acceleration changes from negative to positive and back periodically. Perhaps it’s possible to add a bit more detailed explanation about the data? I’m also a bit confused about the differences between Distance, PathDist and X.
Honestly, I know nothing and can’t answer. We just asked the timing firm for the raw data. They are really busy at peak season, but hopefully when they get a bit more time they can clarify and we will share here.
The vendor’s site suggests they would be worn on the torso in (e.g.) American Football.
https://isolynx.finishlynx.com/
I believe they are measuring the (x,y) positions relative to a reference point / receiver in the stadium somewhere, and that the “distance around the 400m track”, velocity and acceleration are derived from this.
Having been involved with timing at the Night of 10,000 PBs for a few years (where we only got a chip-timing data point every 200m), I can say that there’s always a degree of real world “mess”, especially in the first few seconds. There may be false readings around start and after the finish, races where it didn’t work, and outlying values you can’t trust. That’s part of your challenge ;-).
Is the dataset for track events only, or are there any data for field events? I couldn’t see any entries for field events.
I am currently looking for a dataset that includes fields like the following for triple jump analysis:
- Official distance
- Wind speed
- Approach speed (split into 11–6 m, 6–1 m, and average)
- Toe-board distance
- Effective distance
- Hop, Step, Jump: distances and percentages
I found two papers with tables that include this data, so if a full dataset isn’t available, I can build one based on the samples provided in these papers.
Regrettably the data we have seems only to be for track events. I honestly don’t know if they set it up for field events in Rome 2024. Sorry