- Client: InTheRace
- Services Provided: Data Science & Machine Learning
InTheRace – datadriven performance optimalisation
Can you improve the performance of (professional) cyclists by gathering enough data and crunching it real-time to a level where it becomes an instrument for the team’s coach and medical staff?
Wearables & sensor data
First and foremost, the people at InTheRace had to find a reliable device that could collect the data we needed in our models. Accuracy of the different measurements was a crucial point of attention here. Because we had to collect different parameters (location, speed, cadence, acceleration, heart rate, power, …) it was of great importance to select a device with a sufficiently open character that allowed us to link different types of sensors. In the end InTheRace selected Quarq. Their ‘Qollector’ could easily be connected to existing sensors such as heart rate monitors and power meters on the riders and their bikes, and it also allows us to connect new sensors in the future (for example, air loss sensors on the tires or blood sugar level sensors on the riders).
Collecting data is of course only the beginning. The first tests showed that we could collect an incredible amount of data during a training ride or race, which could give us valuable insights into the performance of a team and the individual riders, up to the second. With these data points, we have started to build a system that should enable near-real time performance optimization.
Convert data to insights
By developing a number of models in R we have started to build insights in the collected dataset. We then also developed a custom dashboard to visualize these insights in a user-friendly way for an often non-technical target audience. In this way, riders and their different parameters can be tracked and monitored live. In order to enable deeper insights into the raw data for end users after the training or race, we are using a combination of PowerBI and Shiny.
Through our dashboard, team leaders receive extremely valuable information almost real-time. In this way they can track and coach riders during the race or training. The riders themselves get a much better view of their performance. So-called dark data, which until recently was not used, is now very useful for riders as well as team leaders and coaches to give feedback during the race on the performance of a rider or to subsequently draw up appropriate training schedules and to structurally improve performance.