I live at 2100 meters. Altitude can skew fitness signatures for those who live high and occasionally train/race low. Each year I get to spent a week or two training (and sometimes racing) at or near sea-level. I am a strong-responder to altitude, so my fitness signature timeline over the last 4 years has “spikes/breakthroughs” at each of my trips of sea-level. The signatures regress to my “normal” high-altitude values over the next several months. This would seem like a great problem for the machine learning tools that sit underneath Xert. Given several years of user power data across a wide range of elevations, one could build a user-specific compensation for altitude. (Note, in my previous data/analysis performance systems I would remove these power files from the system to prevent the skewing of my profile. For obvious reasons, I never liked this solution.)
Great idea Todd. There have been several requests on this front and we have it in our pipeline to address this. It requires quite a bit of testing to get right and we’ll need to dedicate considerable resources for this. Soon …