100 Days of Summer

My name is Konstantin Augemberg and some of you may know me as the Measured Me guy and basically in this presentation I’m going to talk about the self-tracking experiment which I conducted tis summer which lasted for about 100 days.

About a year ago I started a personal quest in which I’m trying to develop my own system for self-tracking, lifelogging, and self-experimentation. If you use analogies from technology business you can think of a self-search engine or Google for your personal life. Building this system will enable me to do a lot of interesting things with my life data including testing and validating diets and exercises as well as discovering meaningful patterns as they emerge over the course of my life.

In order to do so the system must meet certain criteria and the most important one is to be able to capture all the most important, and essential aspects of my everyday life using just a handful of numbers and categories.

It should also have a certain purpose behind it; a meaningful purpose and this purpose is to track myself from the perspective of wellbeing. And this is the most generic and comprehensive definition of wellbeing I found in a dictionary, and I though that’s great, it’s going to be very easy and simple, and straightforward.

But then I started drilling down into the concept of wellbeing and I realized how complex and multidimensional it is. So I ended up with this monstrosity, and essentially eventually I would like to build a system that not only tracks my indicators of wellbeing but also the drivers of wellbeing, the drivers of wellbeing in everyday life settings.

I have to start somewhere so this past year I’ve been experimenting with different methods and ways to quantify and track on a continual basis with two components of the system which are called being well and existential wellbeing.

The being well component basically captures the most important and essential states of my body, mind, and psyche and in its current version it can measure seven metrics; physical health, physical energy, stress, positivity and intensity of my emotions, my mental alertness, and my executive cognitive functioning.

And the existential wellbeing component captures more abstract qualities of life like happiness and life satisfaction.

So what I did this summer was I took this metrics for a long test drive, for tracking them continuously of a 100 days in a row, several times in a day and then I looked a the data to see how well they performed and what information I can get out of it.

So the first thing I looked at was how long does it take to measure and track each of these variables, and as you can see most of them easily pass the test, with the exception for cognitive test because they do take some time to run and I also looked at the amount of missing data as a proxy measure of difficulty of tracking, because the idea was if I intend to skip or forget to measure something that means the measurement may not be cognitive enough.

Then I started looking at the data and analyzing a lot of interesting things about it. one of the first things I learned was that life data is not stationery. So you cannot really apply statistical measurements to it and first you have to develop trends. So it may affect correlation so it is important to trend data before analyzing it.

The second thing I learned is I essentially live two different lives. If you look at the data collected during the work days and compared the data during the aberrations and weekends, you’ll see a lot of distribution and statistical patterns. So it is important to analyze them separately.

I also noticed that some indicators are relatively stable throughout the day and some tent to fluctuate in stable and meaningful patterns, from the moment I wake up in the morning until the moment I go back to bed in the evening. The metrics were the highest and the intra-day variability were energy, stress, intensity of emotion and mental alertness, which is not necessarily a bad thing because I can also quantify those fluctuations.

I can basically use those new numbers to direct new metrics, new additional metrics that could be used to analyze my life and help me understand better. For example I can compute how soon do I get tired throughout the day by simply fitting the slope linear regression model. I can also look at the dispersion of my emotional scores throughout the day and quantify my mood swings and emotional stability, so it’s not necessarily a bad thing.

I also tested for redundancy by looking at the correlations among the metrics, and what I concluded is that my indicators are essentially unique and relative independent from each other, and on the right side of the metrics you can see correlations for the data collected on workdays, on the left side correlations for the data off workdays.

You can see the patterns are relatively stable and also the correlations are pretty much meaningful as you would expect to see something like this.

I also looked at the relationship with the lifestyle factors like sleep, physical activity and also the time I spent on activities like working, creating stock, and spending time with loved ones and there was some meaningful correlations there as well.

In terms of temporal patterns I learned that I tend to be more happy and content with my life on Saturdays which is not a surprise but also on Wednesdays, and least happy and least satisfied on Tuesdays and I’m still trying to figure out why is that.

I also use this data to test the famous Biorhytms Theory to see if there are any cyclical patterns in my emotional, cognitive and physical states and I didn’t find any patterns so far but I will continue testing.

In terms of cosmic effects in terms of weather effects, for example I always thought that rain makes me blue, but the data didn’t show anything. However, hot weather did have a positive impact on my stress and on my emotional states.

You can relate much more interesting things with it, I call it life mining, basically looking for different patterns and exploring data, slicing and dicing data from different perspectives. For example what I did I fit these indicators and K-means cluster analysis model and I came up with a specification system. So any given day in my life was this summer can fall into one of these three categories; it could be a good day, it could be a bad day, it could be a slow day. And the probability of this happening is shown on this pie-chart.

So this basically concludes the summer phase of the experiment and I’m kind of preparing for the next phase which I call Quantifies Winter, and it’s going to last four months this time. During this phase I’m going to improve certain metrics like health and incorporate more biometrics for example, cognitive skills. And I would like to add new metrics like sleep quality, creativity, productivity activity. and now I have some data to back up the idea of maybe measuring some things less frequently during the day.

You can follow my experiment on my blog, measuredme dot com and you can also track my life data with me by looking at my live stream which is also there. the project and data is open source, so you can send questions and download the information and email me and I will send them to you.

Thank you very much.