Analytics is awesome, period. Data, graphs and charts that depict certain parameter readings all tell a story – regarding when a parameter might have been influenced by more interconnected stimuli. Gain the understanding of a graph, and one can quite essentially predict the future!
At Emberify, we hold a special place of interest for personal analytics. Data from wearables and trackers – when correlated and intersected with each other – all depict certain trend patterns when it comes to the functioning of the human body. What might seem a somewhat abstract thought is now turning out to be a reality: we’re gaining a perspective on how we function, and what influences us. And this forms the underlying notion behind the Quantified Self movement, at the end of the day.
So in this article, I take a look at personal analytics and informatics. I try and see how big data and machine learning play their part in bringing numbers to life, and I also talk about how some big names are popularising the practice of personal analytics.
EASING IN: PERSONAL ANALYTICS?
Now when we come across the term analytics, it usually applies to organisations. Big companies use a plethora of organisational analytical tools to assess what makes their workforce tick – be it in the form of motivational factors conducive to the productivity of the employees, or simply how well the company is doing.
I’ve actually written another article on the analytics tools that are used in organisations, and in the article I explain the various types of analytics : there’s descriptive, diagnostic, predictive and prescriptive.
While the above types of analytics definitely apply to organisations and other structures, it isn’t really that difficult to apply the same logics and concepts involved to our lives. And that is where the combined prowess of sensors, trackers and the Quantified Self come effectively into play.
Think about it. A descriptive explanation of what happened through the day – through readings from the wearable fitness tracker strapped on to your wrist. Some days you run five miles, some days you run two.
Why the erratic stats? Enter the diagnostic type of analytics. And this is actually interesting: through correlation with data from other trackers, one can gain a perspective on how things went. A late dinner would imply slight laziness in the morning when it comes to the morning jog. Lack of sleep would indicate less miles run. You get the gist.
And finally, predictive analytics. Hardcore lifeloggers have been known to avoid, avert and at times overcome diseases – mainly because they took care in keeping track of all their vital stats. This is one area I’ll be talking about more in detail later on in this article.
Of course, with prescriptive analytics, things are slightly different in the use cases between organisations and people. While organisations can effectively prescribe plans after the analysis is complete, in personal analytics prescriptions can happen parallel to the other stages.
And if you think about it, the Quantified Self movement is inherently an extension of Big Data. Processes such as Data Collection, Processing and Analysis invariably apply to both the concepts. There’s cloud computing in the mix too (for consumer processing). Software structures such as predictive modelling, natural language processing and the associated machine learning algorithms all play their part when it comes to analysing the large sets of otherwise disjointed and heterogeneous data, in order to make cognitive sense of all the numbers.
All this in essence forms the foundation of personal analytics. Big data and data science, when applied to the Quantified Self, forms a two way communication channel – a biometric variation directly influences the readings, which further translates to real time recommendations.
NOVELTY TO KNOWLEDGE
Through the Quantified Self movement, mind boggling amounts of data are generated. The most notable example, just to gain a perspective on how much data we’re talking about, can be gathered if one studies the Feltron reports.
Initially, all the amount of data can seem to be a novelty, or a mere gimmick. Once a person gets the data regarding the steps walked, for example, the initial reaction is a build up of interest. When I first used my Mi Band, and saw the number of hours I slept, the first thing I said was, “This tiny thing measures all that? Pretty cool.”
The trick lies in channelling this novelty into knowledge. The interest needs to be built up into an intricate system composed of the user, the wearable, the data crunching algorithm and the application ecosystem. It is only then that the data measured will be of any quantifiable use.
And with the power of big data, a lot of the stages that go into converting novelty to knowledge become easier. There’s characterisation, correlation, pattern recognition, which further lead on to prediction of events.
Coming to prediction, there’s this really cool concept which I came across in a paper by Melanie Swan (here’s the link to it), in which she states that cardiac events can be predicted two weeks ahead of time. The process comprises of two main stages, where Phase I involves collecting, storing, processing and analysing the data along with aid from compression and search algorithms. This results in the identification of event triggers. Phase II involves predictions and intervention with low false positives.
Now imagine, if we can predict heart diseases two weeks before they occur, the chances of deaths and complications would reduce greatly. This would in a way pave the path for better lifestyles and more quantified readings of health parameters. All this is ultimately possible, if the novelty is converted into knowledge.
STEPHEN WOLFRAM’S TAKE
Stephen Wolfram is the genius behind tools such as Wolfram Alpha, and he is perhaps one of the most paramount examples when it comes to tracking daily activities and logging them (aka the Quantified Self).
He has been the foremost when it comes to personal analytics, and his search engine Wolfram Alpha and his (now rendered defunct, owing to Facebook’s policies) Personal Analytics for Facebook platform do result in churning out interesting (and informative) information which span out across multiple parameters.
I was reading an interview of Stephen Wolfram on MIT Technology Review (link can be found here), and in the interview Wolfram nicely describes what the big applications might be when it comes to personal analytics.
He describes three major areas in the field – namely augmented memory, where the need to rely on data collected through software algorithms supersedes the need to remember it; preemptive information delivery, which brings about the relation between a person’s history and correlating the current scenarios with it; and finally personal analytics for self improvement.
In his own words regarding the technologies needed to perform personal analytics on a larger scale,
“It’s data science and the whole cluster of technologies that come with that. Then it’s having computational knowledge about the world and being able to make queries in natural language.
Then you need to sense things about the world, whether it’s with sensors or being able to do visual recognition to know what one is seeing.
Then the final thing is just all the plumbing infrastructure to get all of these devices to communicate and feed their information to a place where one can do analysis.”
– Stephen Wolfram
Keeping aside fancy hardware and algorithms for some time, even a deep analysis of the words that we use on Facebook can be enough to tell us about ourselves. For example, here is an analysis of the words that I used on Facebook most often in 2015. The ones that are the largest have been used most often.
The analysis was done mainly on the basis of the posts that I shared on my public feed, and in this regard you can see some obvious trends. Besides the most obvious three words, words such as ‘quantified’, or ‘overview’, or ‘android’ and ‘discussing’ tend to stand out.
Here’s what I can gather on the basis of just this image: I haven’t been sharing a lot of stuff besides most of my articles on my public feed, and I also have been deeply involved with context and the Quantified Self this year. This brings into question how we can tap into data from other sources, and the prospects are quite amazing, really. Think about it: gathering data from email, chats, phone calls and coming to conclusions regarding what you have been discussing throughout the year. Unbelievable stuff, right there.
All in all, personal analytics might just be the perfect combination of cognitive intelligence, data analysis tools, machine learning algorithms and hardware technology to bring about a quantifiable change in oneself.
Again, I’ll invoke some of Melanie Swan’s perspective. There are quite a few information streams within the Quantified Self movement, namely self reported data in the form of health readings, mood journals etc; mobile app data (where Instant comes in!); Wearable device data and finally biosensor data objective metrics.
The trick would be to tap into these information streams and come to suitable conclusions regarding the data collected. Big data presents key advantages in this regard – volume, velocity, veracity, variety and value can be seamlessly applied to the Quantified Self as well.
We’re looking at a stage where the line dividing the concept of Big Data Analytics and the Quantified Self movement is diminishing to the point of being almost non existent. And things can only get more exciting from here on out.
Sign up for our monthly mailing list