I've been collecting personal data on various Fitbit devices for the past five years. During this time period, I've been a high-end college athlete, a washed-up guy sitting at a desk all day, and everything in between. I never collected this data with the intention of doing a longitudinal study on myself, but this opportunity is too good to pass up. There are many natural experiments contained in the data: vastly different bedtimes, variations in type and intensity of exercise, periods of drinking/not drinking coffee, etc.
I wanted to do this study for a couple reasons. First, I've always been a super competitive and driven person, and the opportunity to improve my life and my health is something I will always want to focus on. Secondly, I've been developing analytical skills for many years now, and I think this is a perfect dataset to demonstrate a wide range of capabilities, including:
- cleaning and transforming messy data
- engineering features and bringing in data that does not seem to be related on its surface
- gleaning nonobvious insights based on subject matter expertise (health, exercise, sleep)
- answering interesting questions
