TL;DR: I'm doing a fairly basic project which involves exercise. It seems that descriptive statistics and basic data vis (ex: line graph) would be most appropriate for this project, but I wonder if you have any recommendations for analyses.

For this project, I am performing the same set of 15 single-joint exercises each week (we'll call these "Exercises"). Every 4 weeks, I'm performing 3 different multi-joint exercises (we'll call these "Lifts"). My goals are to:

  1. Track my progress (strength gains) from the Exercises.
  2. See if my rate of progress on the Exercises roughly matches my rate of progress on the Lifts, which would possibly indicate some sort of transfer effect.
  3. Explore relationships between strength gains measured during each workout and other factors (including sleep quality, stress)

My hypothesis is that my strength gains from the Exercises I perform will transfer well to the strength gains from the Lifts that I perform.

The design is a bit janky (and it's more of a correlational study design than an experimental design), but this project is mostly just to demonstrate data collection and basic data analysis concerns. However, I'd like to make the analysis as interesting as possible. So far, it seems that basic descriptives are the best way to analyze this data, with something like a line plot for the gains on the Exercises and the Lifts.

Do you have any suggestions for analyzes that could be performed?

The variables I'm tracking are:

  1. Weight, repetitions, and sets for each Exercise (15 in total, each exercise performed weekly)
  2. Weight, repetitions, and sets for each Lift (3 in total, each Lift performed once every month)
  3. Sleep quality for the night before each workout
  4. Energy before the workout and after the workout
  5. Stress levels before the workout and after the workout
  6. Level of physical fatigue before the workout
  7. Weight before the workout
  8. Heart rate (before the workout and once during the workout)
  9. Exertion of the workout (rated after the workout)



1 Answer 1


I'd start with PCA to see the main correlations and maybe clustering. With the PCA you should be able to find nice sub-spaces, which you can plot and see, which parameters correlate the most/least. Clustering might help you identify "growth spurts". You might also try some linear regression to predict weight repetitions given the other parameters. Here I would be careful though, as the best predictor of performance will probably be the implicit 10th variable: time. You might want to tweak the regression problem to predict changes in performance given sleep, stress etc. In general, depending on what you are looking for, it might make sense to consider "change in repetitions" rather than the repetitions themselves.

I hope this helps.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.