Stop calling analysts 'data scientists'

Tweet
Facebook
LinkedIn
Send to

I disagreed with Mark Ridley, chief information officer of recruitment website reed.co.uk, when he suggested that companies who are struggling to find data scientists should look at sociology and economics graduates.

Why?

Over the past few years, data scientists have been talked up as being absolutely vital to business success, and their relative rarity means they can easily command six-figure salaries.

Such is the hype around the role that the term 'data scientist' is now overused and abused. It's often applied to anyone who has to handle data in their work. When Computing asked organisations whether or not they employ data scientists, the most popular answer was: "Yes, but we call them data analysts". This is doing a disservice to the real data scientists who are out there.

Having spoken to data scientists and industry experts over the past few years, it seems the general consensus is that while many companies have business analysts with a ‘data scientist' title, only a handful of them are actually data scientists.

And so I believe Ridley has a point when he says that those with sociology or economics degrees could indeed be useful from a data analysis point of view. But I disagree that this necessarily makes them a data scientist.

As Birger Thorburn, chief technology officer at Experian's Decision Analytics division, notes, "using data scientists to describe all of them is not always a great fit".

This is because data scientists have an even more specialist skillset than business analysts.

"A business analyst can code a little SQL, use some visualisation, use a classic business intelligence tool, but it's very simple queries and standard report focused, and that's different to what we're talking about with data scientists, analysts and statisticians - people that can get really deep into the code," said Bill Franks, chief analytics officer at Teradata.

"It's the deeper skillset as opposed to the general one that we're talking about," he added.

This deeper skillset is best reflected by former CIO of Credit Suisse investment bank Nigel Faulkner.

"When you look at the skills data scientists have, some parts [of the role] are computer scientists, some part statistician and mathematician, some part economist and some part of domain experience as well. So the idea that you can encapsulate that all in one person seems pretty unlikely when you look at the breadth of the field of our investment bank, let alone the broader remit of Credit Suisse," he said.

That isn't to say that Ridley hasn't had success in taking people straight from social sciences within academia to fill data science roles, but to be an actual data scientist you really do need a background in computer science, statistics or mathematics.

When I asked Jody Porrazzo, a data scientist at multinational media company UBM, what attributes an aspiring data scientist needs, she said a background in data analytics rather than data management was preferable.

"You can come in from a data management perspective with a good idea of how data should flow through the enterprise and then study the analytics and statistics, but I don't think that's the ideal way," she said.

"The ideal way for a data scientist to prepare is to have a great background in statistics, predictive models and business analytics and then to understand how to build out a data programme and understand the importance of data management techniques [later]," she added.

Essentially, a data scientist will approach data in a more exploratory fashion, and will need that data analytics background to be able to uncover correlations and trends for further analysis, or to come up with interesting new findings that no one has thought of before.

But as Gianmario Spacagna, a data scientist for customer and retail banking at Barclays suggested, if data scientists just do analytics, scripting and make models which don't go into production, then a company won't profit.

"It's not a playground. It is not academic. The company wants to make money and you have to solve a problem," he said, adding that companies will stop hiring data scientists when they realise that the majority of them do not bring value.

The trouble is that while many so-called 'data scientists' have data analysis and data management skills, they lack the necessary business acumen to deliver true business value.

Meanwhile, companies are throwing money at data scientists without even knowing what to do with them, according to Mohammed Chaara, director of the Customer Insight Center of Excellence, Strategy & Analytics at Lenovo.

It's because of the hype around big data and data science that companies have been relabelling analysts as data scientists. But data science is a step above data analysis; it isn't the same thing. It's already hard enough to find the best business analysts; applying the ‘data scientist' tag to all of them only makes it harder still to spot the real gems.