What do you do with your data? (How do you gain insight?)

I would like to know how you people engage with your data! I’m trying to get more out of my data myself.

So far, this is what I know:

  • Set daily goals and check your data to see if you meet them
  • Just monitoring the data on a daily/weekly/monthly basis
  • Do experiments where you establish a baseline and then make a change and see if there was any changes in your data
  • Finding correlations between already tracked data

Is this all there is to it?
Or is there more ways to gain clarity from your data?
Has anyone tried using machine learning to gain insight from their data?
Have you tried other statistical methods than finding correlations?

I’ve been on a long quest to find a way to make correlations in my data. I use certain stress numbers to monitor for over-training. Calories in/out. Exist.io has been somewhat helpful recently. Because I have so much data, I know to investigate abnormalities.

Why has exist.io only been somewhat helpful recently? What has been missing?
What kind of abnormalities are you searching for? :slight_smile:

Hi Lars,
You covered a good set of points there! I would add to your list the clarity gained by comparing your data to someone else’s. Imagine you know a group of people who collect the same data as yours - and data and results can be reliably compared). Of course this also hits on some of the inherent limitations of self-tracking (e.g. lack of standard approaches, fragmentation in tools, privacy).


This probably isn’t exactly what you were intending, but I’ve found data has been extremely helpful in many situations when interacting with external people. I’m severely chronically ill/disabled and I mostly track symptoms, and I’ve been able to use my data both to communicate better with doctors and also to successfully argue my case for disability services to fund certain things that they aren’t previously willing to fund because the don’t believe the value of.

I’ve got up to 6 years of daily tracking of some symptoms now, so that’s a LOT of data which does a very good job of telling me whether the difference in a symptom’s severity is within the range of how it’s varied in the past or if the change since I started a new treatment looks new. Unfortunately I don’t have the math required to actually establish it formally, but I put a rolling (usually) 7-day average on my symptom graphs (most symptoms are recorded on a 0-3 scale) and it’s surprisingly visible by eye whether something “looks” new or not. I have no idea how valid this is but I’m sure it is at least better than what memory would dredge up with no tracking at all, and it’s the best I can do for now.