Sleep Data Analysis

In 2017, I bought a Fitbit with the goal of improving my sleep. I wasn't necessarily looking for something to help me regulate my sleep, as much as I was looking for a way to track it. The following is a look at data I collected over the course of 6 months (138 nights)!

The natural place to start my review was plotting the distributions of the sleep data. These graphs give a good summary and capture the core statistics. The Fitbit categorizes various stages of sleep by tracking heart rate and sensing movement, it manages to classify sleep into: light, REM, deep, and awake. In the graphs below, "Minutes Asleep" is the sum of light, REM, and deep sleep, while "Time in Bed" is essentially the sum of all the categories. After exporting the data, I used the two variables to create "Sleep Efficiency"

On average, I get 437 minutes (7h17m) of sleep per night while I spend 516 minutes in bed (8h36m). Naturally, minutes asleep is positively correlated to time in bed. About 81% of the time I manage to sleep more than 400 minutes (6h40m), which is good to see considering 7-9 hours is generally the recommended sleep amount. Inspecting the left tail end of "Minutes Asleep" shows that the only time I get less than 350 minutes of sleep is when I have an early flight to catch or when I am going climbing/hiking outdoors.

My sleep efficiency is on average 85% and seems normally distributed. On any given night, if I want 7.5 hours of actual sleep, I should aim to go to bed 8.5-9 hours before my alarm clock is set.

I also plotted the data against the day of the week, however I excluded travel days as they were not representative of a typical day. The graph shows that on average I get the most sleep on Saturday's (Friday nights) and Sunday's (Saturday nights), this is expected since I like to sleep in on the weekends. Additionally, weekends see the most variance in how much I sleep, presumably because there's a mix of nights where I am catching up on sleep debt, and nights where I go out and go to sleep very late. There seems to be a downward trend in how much sleep i get during the weekdays, particularly from Monday-Wednesday, since I wake up at the same time for work everyday this mostly means I tend to go to bed later as the week progresses.

The last graph I created is simply a summary of my time asleep. There is data missing as sometimes I forget to wear my Fitbit to sleep. Areas in gray denote vacation or travel