Some time ago, curious about why some nights I slept longer than others, I started logging bed and rising times, along with diet and exercise. Finally a few weeks ago I started poking around my linux distribution for statistics software. The package information being minimal, I started poking around the internet, and eventually settled on R, which (perhaps unsurprisingly for me) is a kind of statistical programming language. In addition to coming with plotting functions and standard statistical functions, it intrinsically supports vector arithmetic. The main problem is all the documentation assumes the audience are trained in statistics, so I can only get basic information at this point. I’m working my way through a statistics textbook that I grabbed out of an end of year ‘free’ bin at the Cornell bookstore.

The early results show a possible seasonal effect (Until I collect multiple years of data, finding sunrise and sunset data will be my best bet to confirm that) The best predictor of length of sleep is bed time. Adjusting for that, there is a curious spike on Saturday nights. The couple of diet classes I’ve entered so far show little correspondence, although the soy milk may have had a small effect. Exercise shows no short-term correspondence (all analysis has been night-after so far)

Hopefully further improvement will come from extra data sets and improved statistical methods.

Comments are closed.