How much of data science is witchcraft?

Trying to explain what Data Science is in layman’s terms can be difficult. People are intrigued by the title, but “what does it mean, exactly?”

Invariably, the mystery deepens as puzzlers wrestle with the idea of a data and science mash-up. And as conversations continue, the whole thing sounds more and more like witchcraft.

Data scientists work with car manufacturers like Volvo, to design a future where no passenger is ever hurt or killed in an accident again. Think what that would mean. Imagine the impact.

At the same time, data scientists work with Amazon-type companies to ensure that a consumer’s purchase is moved to the nearest distribution centre before they’ve even clicked the buy button. Then delivered in 60 minutes.

Just a couple of examples that spark interest and curiosity among friends. But can data science magic such amazing results from data and analytics… and fresh air?

The missing link

Well, on its own, data is simply numbers and words; digits and characters. Similarly, analytical algorithms are mathematical formulae translated into code. And applying a random algorithm to an anonymous dataset is like fishing in the dark, and the chance of success is about the same.

The fact is, a dataset has to be understood: the business context; the contents; the individual values; the distributions, etc. Knowing the dataset and the particular business challenge helps the data scientist select the perfect algorithm for converting the data into information and actionable insights.

The reason data science can’t be fully automated at the moment is because the granular discussions and decision-making that lead to a workable solution haven’t happened yet. In other words, data scientists are still waiting for the rulebook to be written.

Have you got what it takes to be a data scientist?

Any analyst can follow a data cookbook that provides the ingredients – a specific set of datasets and algorithms – and a straightforward recipe. However, you need a data scientist to create that cookbook in the first place. He, or she, has to think outside-the-box to create something fresh and innovative – from scratch.

Of course, you have to have in-depth, wide-ranging knowledge and experience to write a rulebook. But that’s not all. A data scientist has to be a natural problem solver. In fact, the full palette of skills is hard to pin down. Moreover, it’s nigh-on impossible to teach a good candidate to be great at this.

Either you’ve got it, or you haven’t.

Innovative versus repeatable analytics

Despite businesses having spent years coming to terms with traditional analytics, they’re struggling to find the right people with the right skills for today’s innovative analytics. Teams are having to cope with new datasets, like digital clickstream interactions and IoT readings, which haven’t been exploited before. There are limited tried-and-tested methodologies to get value from these datasets. But to get maximum value, innovation is the only real answer.

The first business challenge is making a clear distinction between repeatable and innovative analytics. Next, innovative thinking is required. Unfortunately, most businesses have limited in-house skills. So, they have two options. Make a concerted effort to attract experienced people who are unique in the way they think (people who may not have followed a conventional path and gained the data scientist title). Or partner with organisations that can provide the necessary skills and experience.

To the uninitiated, data science may seem like witchcraft. But working with an ‘out-of-the-box’ thinker – someone who understands data and knows analytics inside out – just might conjure up the incredible solutions businesses are looking for.

Some of which… maybe, just maybe… they never even dared to dream.

Yasmeen Ahmad, data scientist, Teradata

Photo Credit: Sergey Nivens / Shutterstock