100 Variables the model must consider, GO

TrainerRoad Software
1

What should the AI/ML be considering to really give you those dialled workouts and predictions? I’ve seen quite a few discussions about heat and Whoop lately and it got me thinking.

Add one item (at a time, go wild with ideas), number it.

If we hit 100, @eddie will add them all individually to the backlog :laughing:

I’ll get the ball rolling…

  1. Coffee consumption, rated as only: too much // too little.
  2. Sleep score, fed by the users baby monitors.
  3. Number of local legends the user feels must be protected.
  4. …
2
  1. The amount of cartilage left in your knees
3
  1. Your boss’s mood
4
  1. The Interference Effect of Gardening
5
  1. The relative humidity of your pain cave after various periods of time, 30 mins, 45 mins etc.
6
  1. Drivetrain heart rate balancer, because your heart rate doesn’t drop descending at 0 watts // 120 rpm on the fixed gear pub bike.
7

That it is 10% wrong.

8
  1. The current state of the TR forum
9
  1. The number of items on your DIY list this weekend
10
  1. Your wife and/or toddlers mood
11
  1. How many TR podcasts you have listened to this week
12
  1. @Jonathan Leadville progress against FTP target.
13
  1. Your cognitive capacity that day - as determined by number of guesses to get the daily Wordle
14
  1. The % delta between how you answered the post workout survey and how it actually felt.
15
  1. Number of forum posts the user has read about other user’s AIFTP being wrong.
16
  1. The changes made to the calendar to increase the FTP prediction
  2. The discrepancy between your and your buddys W/kg trends
17
  1. The colour of the socks you plan to wear.
18

Penalty for triathletes wearing no socks

19
  1. Height of socks worn during workout (Is there an AI Sock Height Prediction? Like if I complete enough workouts at prescribed level the height of my socks should increase ?)
20

Looks like we followed directions thru #18 then we lost it.