Spreadsheets are an excellent learning canvas to get under the hood and to build intuition for how data science algorithms work.
What’s the fuss about data science?
Most times I unlock my phone – I’m not entirely certain whether in five minutes, I’ll be watching a video, listening to a song, playing a game, getting my news feed, shopping or networking. That’s because, beyond consuming these online services, unbeknownst to me, I’m also interacting with several machine learning algorithms vying, at the same time, to nudge my attention and my choices.
Few domains in our personal or professional lives remain unaffected by machine learning and data science. Businesses have thrived by integrating data science into operational strategy. Others have lost ground to competitors for ignoring it’s latest advances.
Algorithms now decide our credit score, which patients receive medical care, and which families get access to stable housing. This hidden web of automated systems can trap people in poverty.
— MIT Technology Review
While it’s not to say that my aim or yours should be to become a data science engineer overnight, good understanding of data science is a matter of remaining relevant at workplace and, to a large extent, about being aware of how our choices and decisions are influenced by machines.
On one extreme, the dreadful god forsaken year that 2020 was, showed us getting data analytics right can literally be a matter of life and death. On the other end of the spectrum, the novel applications of data science and machine learning might tickle a brain curious not only about how it all works but also about why it works at all and when might it all go wrong.
But why spreadsheets? Familiarity is why.
Right at the outset, if you want to, with your own bare hands – code a machine learning solution and scale it up for use by thousands of users, then ignore spreadsheets by all means.
But if you are an executive sponsoring or leading a machine learning project, then spreadsheet, specifically MS Excel, is an unparalleled learning platform because it is, in spite of all its shortcomings, the most ubiquitous and familiar analytics platform. It doesn’t hurt that excel is a very visual analytics tool needing very little or no code. Often through its tedious formulas and admittedly inefficient implementations, Excel will lay bare not only how an algorithm functions, but also which parameters affect performance and when things don’t work. It’s the perfect DIY platform to get your hands dirty.
Finally, why delve into data science if I’ll never write a line of code?
It is common knowledge among experts that a majority of machine learning products have little or nothing to do with machine learning; products labeled “AI enabled” is often an unreasonable exaggeration if not simply a deception. That trend isn’t a coincidence in a world where Investors, Business executives and users have little or no hands-on understanding of data science beyond a crude and misplaced assumption that machine learning and artificial intelligence is akin to a black-box solution. In its worst form, machine learning projects become an expedition of spending millions of dollars only to build a product which reinforces management’s biases and instincts. I’ve experienced this first hand.
As a practical matter, understanding the mechanics of data science could mean you avoid wasting time and resources on projects to solve problems which haven’t been conquered by the most cutting-edge data science models. It could also mean unlocking opportunities to solve problems previously considered unsolvable. But most importantly, it could mean you separate problems needing Machine Learning from those which can be solved with rudimentary business rule tables.
But, the reason I really want to explore data science and machine learning on a spreadsheet canvas is the same reason I like any DIY projects. Its a lot of fun!
* The post Learning Science vs Tools first appeared on continuoous.com