Getting Down With Data Science
Making the most of your data scientist
D/SRUPTION spoke to Jason Mander, Chief Research Officer at market research company GlobalWebIndex, about the role of the data scientist, the unique insights they offer, and what businesses can do to maximise the potential of these valuable employees.
Data is one of the most valuable assets a business can possess. With it, organisations can create new markets, drive productivity and generate much needed growth. In today’s world, almost everything we do creates data, leading to a wealth of potential business opportunities… But getting hold of data is only part of the story. Once an organisation collects information, it needs to work out what to do with it. This is where the data scientist comes in very handy indeed…
Umbrella term, important job
Uniquely placed to make sense of big data and suggest business avenues accordingly, the data scientist is vital to the large organisations of today. In general terms, the data scientist combines the skills of a mathematician, analyst and computer scientist. However, the role varies in line with the requirements of specific companies, as Jason Mander explains.
“Data science is an umbrella term describing individuals whose job is the art and science of deriving value from data,” he says. “Outputs from data scientists can take many forms and flavours, from the delivery of intelligent features integrated into business products (think Amazon and Netflix recommendation engines…) to mathematical modelling and visualising information. Underlying these different applications is a skill set encompassing everything from machine and deep learning to statistics, computer programming and product design.”
With such a diverse range of tasks and skills falling under the data science umbrella, Mander notes that uncertainty arises around the term. In fact, what a data scientist actually does is often poorly understood within the business community. The role can be similar to better known jobs such as business analyst, statistician and chief data officer, but should not be confused with them.
“Unlike a data scientist, a business analyst typically does not use machine learning or predictive modelling,” Mander says. “Statisticians are typically not as strong in computer programming as data scientists, and the role of chief data officer is an even broader term that encompasses all issues pertaining to data, not just analysis and prediction but also data governance, processing and security.”
A work in progress
As with other roles at the forefront of disruption, the data scientist has seen rapid change over the past few years. This demands the right kind of attitude from data science employees. Keeping up with technological developments is a must, as is opening up their remit to wider members of the business.
“At one end of the spectrum,” says Mander, “there’s been a focus on automating a large part of the data scientist’s toolkit and workflow, thereby allowing individuals who are not trained in topics such as machine learning to effectively deploy machine learning models. This has given rise to the so called ‘citizen data scientist‘ – someone who is not an expert at the algorithmic or mathematical level, but who can execute and interpret the results of an analysis and what it means for a business.”
“At the other end,” he continues, “data scientists have needed to embrace sophisticated new techniques such as deep learning. Thanks to improved infrastructure and the growing availability of open source tools, these techniques have pushed further into the business mainstream and have raised the bar in terms of the methods and the approaches that data scientists need to deploy. For today’s data scientists, keeping up to date with skills and techniques is more important than ever before.”
Historically, business data has always been a closely guarded secret. In the past, fears over competition and siloed IT infrastructures meant that data was typically held by an organisation’s IT department, with restricted access rights dictating their use. Unfortunately, not only did this mean that competitors could not get access to valuable data, it also made it difficult for members of the business to make use of this information. When data exists in complex, inefficient systems, it is hard for an organisation to make quick, data driven decisions.
In recent years, businesses are increasingly looking to open up their data, in line with the concept of ‘data democratisation’. As Mander states, this involves an organisation giving relatively open access to its anonymous, non sensitive data. Data sets that would once have been guarded have become available to any individual to access – hence the term democratisation – in a move towards open data.
“The benefits of this to business are many fold,” Mander says. “Quite apart from promoting a sense of transparency, it allows individuals to analyse data sets to see what value they can add and what insights they can find. It’s the YouTube principle in action: just as no individual or single team could ever produce the wealth of videos uploaded to that platform by its huge userbase, so the collective efforts of countless data scientists are likely to produce results which are richer or more diverse than if left to a single team alone.”
Although this kind of data strategy presents opportunities, it does come with a warning tag attached.
“For a business this demands a culture shift around data stewardship and data transparency and scrutiny,” says Mander. “Equally, it lowers the barrier for competitors to derive insights by analysing data that is widely available. There’s also always the obvious danger that people who are unqualified might misinterpret or misuse a data set.”
Getting the most out of your most valuable asset
Although data can be described as a business’s most valuable asset, it is only as good as the people who make sense of it. This means that, along with hiring data scientists with the relevant skills – a task complicated by the digital talent gap – it is crucial for organisations to manage them effectively. For Mander, this involves giving data science teams the vision and purpose they need to guide business decisions.
“Context is king for the data scientist,” he says. “No matter how talented and skilled they are, any model they build to describe or predict data will only be truly powerful if they understand what the business wants to achieve.”
“They need to be equipped with the business vision, a description of what success looks like for a particular analysis or feature, as well as with priorities. In fact,” he adds, “much of the guidance that is typically needed for an engineering or development team applies here too. You need to bridge the gap between the data science team and the rest of the business, despite the fact that these two entities tend to speak very different languages.”
Along with communication, an additional challenge for organisations is the high degree of uncertainty that surrounds some data science projects when they begin.
“For example,” says Mander, “a project where the goal is to automate the proactive servicing of factory components by predicting machine faults ahead of time needs a certain predictive accuracy to be cost efficient. This may or may not be achievable using data on past faults that the factory has collected. In short, just because a data scientist manages to produce a valid output, it doesn’t automatically mean that it’s commercially viable.”
The optimisation of data can only occur, then, if an organisation has someone to understand it, turn it into actionable insights, and align it with a company vision and goals. Add in expertise in fields such as machine learning, modelling and data visualisation, and it’s no wonder data scientists are so indispensable to large businesses today.
For more insights from industry experts sign up to our free weekly newsletter.