I bought my Fitbit Flex from Costco sometime around 2:00 PM on January 4th, 2015. I had been holding out for the Charge HR model - which ironically was released the next day - but ultimately decided that having continuous heart rate measurements wasn’t worth the extra $50. After all I am certainly capable of taking my own heart rate.

Its been about 7 months so I thought I would take a look at the data my Fitbit has collected so far. The Flex tracks steps, distance, “active minutes”, and calories burned. Other models track flights of stairs and heart rate. Fitbit allows users to download their own data in an aggregated form at a daily level and offers downloads at a higher temporal resolution as part of their premium membership. I don’t pay for their premium service but fortunately intraday data is available through the API that generates graphs on their dashboard page and there is a package on CRAN and Github written to take advantage of this. So accessing your own Fitbit data is actually quite easy.

Steps and distance

My steps and distance were highly correlated ($p=0.9961$). In fact fitting a quick regression showed that the relationship between the number of steps I take and the distance traveled is a simple linear equation given as $\mathbf{D} = -0.0014297 + 0.0004477 * \mathbf{S}$. It is safe to say that the intercept should be zero (no steps equals no distance) and dropping the intercept from the equation gives a model that performs just as well ($R^2 = 0.991$ without the intercept vs. $R^2 = 0.992$ with the intercept).

The graph below shows the number of steps each day with reference line (white) at 10,000 steps, the default daily step goal, and a loess fit line (yellow) to show the local averages. My daily average is about 10,853 steps (or 4.86 mile), slightly higher than the goal.

There is a slight upward trend during April and May with a return to the mean in June. This is partially driven by a trip to Chicago over Memorial Day weekend and moving further from work (and walking to work less frequently). Two weddings stand out, one in the middle of June and the other the end of July, both exceeding 18,000 steps.

Active minutes

Fitbit uses metabolic equivalents (METs) to calculate active minutes. Active minutes are essentially the number of minutes where the number of steps exceed some threshold, with a default goal of 30 active minutes a day. For other models active minutes are also based on heart rate.

My average is 39.5 active minutes (median is 37) but there is also a good amount of variation. My active minutes are strongly correlated with steps ($p=0.7496$) since more intense activity or consecutive steps are necessary to achieve active minutes. Features identified in the steps graph are also visible in the active minutes graph (below).

There is a very distinct drop in my number of active minutes following Memorial Day weekend and corresponding with moving further from work (and walking to work less frequently). This change in daily active minutes is more dramatic than the change in steps indicating a decline in intensity.

Retrospective

The Flex provides interesting metrics however the models with heart rate monitoring are compelling because they use that information to improve active minutes calculations and estimates of calories burned. My data shows that although I average more than 10,000 steps a day, there is also a large amount of variation day to day. At some point it would be interesting to pull in additional Fitbit data (i.e. sleep, calories burned) and external data (e.g. weight, temperature) and look for general correlations.