I'm not the average user. What I want in a fitness tracker is this: accurate, complete data. In long form:
My fitness tracker would capture every step and moment of sleep, and have a "dump" button that dumps everything it tracks into a tabular form, where my ETL's can pick it up and write what I need into a database.I haven't found a fitness tracker that will do exactly that, but I found one that I thought was pretty good, here's a picture and some pros and cons:
- Always on. Only requires charging once a year, and goes on my wrist, so I literally never have to take it off. My wife uses a different tracker, one that you wear on clothing, and requires weekly charging, so if she was modeling her data, she may be annoyed by this built in error.
- Tracks sleep. Not all fitness trackers track sleep, but this one does; fairly accurately from what I can tell. I like it because it allows me to work on variance reduction in my sleeping patterns. I've also found that hours of sleep is predictive in a few of my models.
- Easy to sync. All I have to do is push a button for a few seconds, and it auto syncs to my phone. Easy.
- Data is accurate. I've done a couple of tests, one wearing a manual pedometer, the other just knowing my stride length and how far I've run. Generally, the Garmin Vivo Fit 2 is within 3% variance of actual steps taken.
- Data created imperfect. To create my database of activity for analysis, I have to create spreadsheets and import them, but the Garmin tool doesn't create great data (sometimes using days of week rather than actual dates, and not giving me a "dump all" option, instead having to take one or four week blocks.
- Doesn't count elevation. My wife's tracker counts elevation in terms of flights of stairs. Because my runs often include a big "hill" component, this would be nice for me.
I really like this device, and I'll continue using it. I think in the future these devices may better integrate with my homebrew data systems, but for now, it is at least a working, accurate system.