Statistical (Sleep) Wizardry

Last week I highlighted a few tools that are on my radar for doing quantified self data analysis. I’ve spent a bit of time playing with Wizard, a Mac-based statistics package, and gained a few insights from it that I’d like to share.

Each day I get emails from a set of recurring Google Calendar events which contain links to personal surveys I set up in November of 2015. The surveys contain a few questions about variables I think affect my sleep, along with basic emotional and mental self-assessments along the lines of “How happy are you feeling?” and “How well are you focusing?”, among others. The data is aggregated into a Google Drive spreadsheet for analysis, and at this point I have a little over 200 entries. My goal in collecting all of this was to help identify factors that promote healthy sleep and allow me to maintain focus for long periods of time.

Today I took Wizard for a spin, throwing the survey data at it and trying to glean some insights. Here are a few things I noticed:

  • Unsurprisingly, there is a significant positive correlation between how tired I am and the amount of coffee I consume.

clojure.lang.LazySeq@29f5b776

  • I focus better on days when I feel well rested, and coffee does not seem to have much of an effect on how well I think I’m focusing. I also feel better emotionally when I'm better rested.

clojure.lang.LazySeq@ecebd446 clojure.lang.LazySeq@ecebd446

  • The number of drinks I had the night before does not seem to affect my sleep below a critical limit - about 3 drinks.
  • Eating “healthy” foods does not seem to affect my sleep very much.
  • Getting a bit of that sweet time with the lady (or solo) has a statistically-significant affect on improving my sleep.
  • So far exercise does not have any significant relationship with sleep quality.

There's definitely some interesting results in there, and I look forward to playing with the tool more. In particular, I want to pull in environmental data and the results from my FitBit to see how they play against my self-reported data.

If you like what I write, you can subscribe to my mailing list to get new posts in your inbox: