How data science changed my brain

Math wiz, not

My dad was a maths professor at a local college for some time and, unfortunately for the both of us, I was not better than his worst student. Throughout high and undergraduate school, I solely pursued arts and humanities to gain degrees in psychology, business ethics and organizational leadership. I was an unemployable pseudo-intellectual.

So I went to graduate school to learn analytics. Within a span of a year, I had become a full-fledged data scientist and gained employment in the data science industry. Now that I've worked in this field for a couple years, I find that my head has been turned all the way around — the way of computers, numbers and models made my intuition got much stronger.

Data science didn't endow me with a new found intuition, but gave my existing intuition structure; that is, a sense of the rules, rates and relationships of what I am intuiting. Navigating your in-born intuition with form and function makes it much more efficient and useful; as an inquisitive person, I found myself asking better questions. It had another effect, however. It also made me feel as though I had peered behind the curtain, saw the wizard, then realized I, too, had a slight tinge of green. And it was paralyzing, because here's the thing: Data science gives you really powerful problem-solving tools, but it’s still on you to ask good questions and make sense of the answers you get. No getting around that. What question should one ask?

Data science word salad

“Data science” itself is an egregiously vague term. It used to be called e-Science, which I think is a much better fitting term because it centers the role of data in the ethos of science and highlights the very new and real potential for science in a distributed digital space.

Anyway, I worked at a company whose likeness is near-synonymous with the entire field of data science and I interfaced with its vague nomenclature a lot. To start, I’d like to break out the mathematical constructs we use.

Statistics

The art and science of making sense from sampled data.

When you learn statistics, you are endowed with techniques to summarize data, formulate and evaluate hypotheses with an acceptable tolerance for reliability, and henceforth make probabilistic predictions. We do this to understand the past, navigate the now and slice into the future. Statistics is more than sufficient for most moola-printing questions.

Calculus

The art and science of unraveling the intricate mechanics of our reality.

Offering capabilities beyond statistics, calculus lets us wield data in more dynamic and complex ways; when the research question involves multiple continuous factors, calculus can help us understand it and see where it might be headed. Put simply, its used to study change.

Calculus is built upon the same mathematical constructs of algebra and trigonometry and uses the concept of derivatives and integrals to give data form and motion. This helps us understand rates of optimization, accumulation, and transformations and is appropriate for dynamics that are continuous or "never end.”

Think: Planetary movement.

Machine learning (or niche AI)

The art and science of self-running optimizations.

Machine learning is, in and of itself, an application of calculus that is designed to react to real-time data. It is the technology behind the algorithms that determine your houses’ Zestimate, the next turn on your Google Maps and your next Tinder date.

Machine learning was always the "fancy" realm of data science and the application I looked most forward to learning. Funnily enough, it was actually a very unsatisfying learning experience — a few lines of code can get your machine "learning." That said, what the machine learns is the most important issue of our time.

And back to me again

Am I a master of my pixelated environment?

Once I got a handle on how data science worked, it dawned on me that Maslow’s hierarchy, in hand with data science, could truly transmogrify real-life decisions big and small into some sort of utopic (or dystopic?) video game rife with handy tooltips and illuminated paths, gently guiding me to that non-arbitrary point of human optimization.

The confidence that data science gave me made me feel as though I truly could, in theory, solve for any definable problem. And here’s where the psychology bits comes in — and it’s lodged somewhere between knowing that you can help yourself and actually helping yourself.

The roles of motivation and belief

Let's take a look at the societal structure that modernized capitalism and can produce outsized wealth for individuals: the stock market. Did you know that 8% of White Americans own 89% of stocks in the US stock market? Bananas.

What the stock market enables is nothing short of a miracle, but I wouldn't buy what I don't understand either; particularly not from an industry whose main cushion is consumer confusion. But being from Hong Kong where there are no incentives to leave credit on the table (its culture is of risk aversion, an orientation towards savings and history of pandering to institutions), I know Americans take credit card perks, tolerable APY saving rates and a generally growing stock market—a bonafide consumer haven—for granted. But lunch isn't free, so...  

So get on with the Gamestop bros and cash in, right? Nah. It's good enough to be able to imagine how the stock market works (with a tolerable level of self confidence).

Armed with data science, I feel OK opting myself out of complete and perfect financial literacy. I am OK relying on available signals, whether that be signals from the market or from cultural / behavioral shifts writ large, which has been a large departure from being avoidant. I do check sources and I do verify, but always as little and as quickly as possible. The point is that accepting that signal, however simple, basic or obvious, helps me wayfind and move the heck on. We should lovingly tell ourselves “that’s above my pay grade" more often.

The Bayesian Brain

I enjoy the occasional metaphysical discussion. I enjoy thinking about philosophy and everything else I’ll never know for sure. And according to Pew, I’m not alone. Apparently 41% of Americans believe in psychics, astrology and/or reincarnation.

What is interesting though, is that religious affiliation (with the exception for Atheism) does not significantly correlate to these 'new age' beliefs.  Put another way, everyone believe in psychics, astrology or reincarnation at similar rates regardless of religious affiliation. Isn’t belief fascinating? Turns out, there are no rules and we don't care.

I remember watching The Secret, a book-turned-movie about ~*magical manifestation*~, and fully experiencing it as a parody. If you haven’t seen the film, The Secret is essentially a continuous B-roll of testimonies convincing you of ~*the secrets*~ efficacy. Turns out, the book-movie could have just as easily been called The Bayesian Brain.

‍

‍

Bayesian statistics, developed out of Bayes’ theorem, describes the probability of an event based on prior knowledge and can integrate new or emergent data to further investigate the parameters at play. I love how uncanny it is that the Bayesian way to express interpreted probability is as degrees of belief.

The Intersection

I know this is all over the place, but it's a work in progress. Bear with me: Consider the fact that, in a year, there are 31.5 million seconds. Now consider that the unconscious mind processes up to 10 million bits of information per second. So, how many thoughts do you have in a year? Ouch, nevermind.

According to Cleveland Clinic, the human mind processes upwards of 70,000 thoughts every day. Some other facts, the brain has:

• 100,000,000,000 neurons
• 100,000,000,000,000 synapses
• 500,000,000,000,000 connections
• Traveling at 300 miles per hour

Now let’s consider one of the worst things that can happen to someone: getting a cancer diagnosis. Imagine you’ve been diagnosed with an incredibly rare form of cancer (only affects 0.5% of the population), you’ve just taken a test that has 95% accuracy (given you do have cancer) and it comes back positive. You almost most certainly have cancer. Right? No. Let’s use the theorem.

(0.95 * 0.05) / (0.95 * 0.05 + 0.095 * 0.1) = 8.3%

Turns out, your real risk for actually having cancer is 8.3%. Cases like this highlight how physiological and safety needs implicate the attainment of self actualization (the cognitive processing of difficult information).

Taking a step back from the anxiety of it all, this number should actually feel much more aligned to what is inherently believable given the incredible rarity of the particular form of cancer (and I’d argue, the aesthetics of it not being a whole number).

I am the machine, learning

In 1983, Carnegie-Mellon and John Hopkins investigated how computer systems could replicate "intelligent behavior."⁸ They were modeling AI after the human mind and describe how the brain efficiently parses information; the implied enormity of computation required to achieve what the human mind does would require something quaintly referred to as "massive parallelization." 40 years later...

Belief is as significant as the placebo and nocebo effects.⁹

‍