Estimate Power without Measuring it: a Machine Learning Application


  • Guillaume Lemaitre Paris-Saclay Center for Data Science, Palaiseau, France
  • Cedric Lemaitre Apeira Technologies, Le Creusot, France


Power is the core metric of performance in cycling. Trainers strive for maximizing it while engineers strive for minimizing its loss by improving equipement and aerodynamic drag. However, costly and specific equipments are required to optimize this process. Here, we show that power can be estimated from side data without the need to acquire power data itself. More specifically, we show a proof of concept that machine learning can serve to estimate metrics with limited equipment or missing data. In this work, we limit the scope at estimating power but aerodynamic drag could be estimated in the same spirit.


Download data is not yet available.



How to Cite

Lemaitre, G., & Lemaitre, C. (2019). Estimate Power without Measuring it: a Machine Learning Application. Journal of Science and Cycling, 7(2), 48-49. Retrieved from

Most read articles by the same author(s)