3 min read

Accidental RAP

I have this pipeline of data - I do a workout, record bits as I go along, log it on Fitocracy…

Fito haven’t added any features in years, in fact they’re barely keeping the server running at this point.

So I grabbed this script that asks the server for all my workout data, which arrives as a giant blob of JSON.

The data on my machine doesn’t need to be the most timely, so I have cron run that script 1ce a day. 1

There’s interviews with the founders of Fitocracy that showed that they distrusted libraries and wanted to code everything from scratch. This really shows in the JSON that the server spits out. I have this little script that converts that JSON into a CSV:

Honestly, the heavy lifting is done by the jsonlite package. The rest is a bit of purrr going up and down the list of lists of lists and keeping only the useful stuff.

Now I have a (massive) csv that’s accurate up to midnight last night. Then the below dashboard sort of grew out of that.

It wasn’t until someone gave a presentation on Reproducible Analytical Pipelines (RAP) that I realised I’d built one just trying to motivate myself to keep lifting heavy stuff.

Dates by PR

name kg reps
Barbell Bench Press 110 1
Barbell Deadlift 190 1
Barbell Squat 180 1
Standing Barbell Shoulder Press (OHP) 70 1

Volume for the year

Workout days for this year

The programs I’ve been working with while running these scripts have focused on pressing things overhead, pressing on a bench, squatting and deadlifting, so that’s why those ones get more attention.

I came across Stronger By Science’s Art & Science of Lifting which confirmed that (as long as the weight isn’t too light) the better workout is generally the one with more volume - the repetitions (reps) times the weight. So I started graphing volume.

The heatmaps of volume for those lifts are kinda inspired by GitHub’s activity graphs.

I installed the ggalt package just so I could draw dumbbell graphs.

Further work

The dashboard gets tweaked fairly often, sometimes adding a table/removing a graph/changing a colour scheme.

Blogging about how I accidentally built a RAP has been on the TODO list for a while…

I want my nightly cron script to pin the latest data with the pins package, which I’ll do by tweaking the json->csv part of the pipeline.

  1. For 1 user running this script 1ce a day, I don’t mind if it’s inefficient in that it takes 10 mins. It’s far more important to me that it works.