Science has conferred a new dignity upon data, in the term "data science". Matin Jouzdani, an associate partner at IBM, explains the collocation as meaning, “Data analysis is increasingly difficult, so the people who do it have been called scientists" in the past couple of years.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
For business leaders, the desire for a data science capability has been strong recently. Why? “They are reacting to hype,” says Debra Logan, analyst at Gartner.
Nevertheless, the desire is there, even if it is merely a sign of a recession-driven need to turn the base metal of data into gold.
"There is still a perception that a data specialist, perhaps a recent statistics graduate, should be parachuted in to an organisation to advise on how to work magic with data. This is flawed thinking. Data is everyone’s business," says Jim Giles, the author of a recent Economist Intelligence Unit survey-based report, entitled Fostering a data-driven culture.
The survey results indicate that the most successful companies optimise the use of data by providing necessary training and promoting the sharing of data across all levels of employees and departments.
The report, sponsored by data visualisation company Tableau Software, argues for a democratisation of data analytics over an excessive focus on the development of a cadre of data scientists.
Data analytics at Lastminute.com
Bill Beckler, chief innovation officer at Lastminute.com, seeks to balance the democratisation of data with the building up of a rigorous data science function.
“We are strong believers in grassroots access to data tools so that data can drive decisions everywhere," he says. "This makes everyone's job easier and more satisfying, since it gives people more responsibility and the tools to ascend higher and higher in achieving their goals. Our data science team is focused on this approach.”
The Lastminute.com data science team was built with an open recruitment process, according to Beckler, which included solving very difficult maths problems that can only be solved by someone who has the skills to handle the challenges the organisation faces.
The team’s remit includes data processing, predictive modeling, marketing optimisation and answering strategic business questions. "Team members are constantly building tools to automate what they can solve manually,” he says.
Excel stops working at a million rows. At that point you need a data team
Bill Beckler, Lastminute.com
The company believes in hiring stars. "To retain stars, you need a constellation. So we keep our standards high and make sure everyone is someone with whom everyone else would love to work. This makes development and retention easy, since morale is high and excellence is contagious,” says Beckler.
Stepping back from his own company, what types of organisation does Beckler think would benefit from having a team of data analytics specialists?
“As soon as you need to make decisions based on a lot of data, you need specialists who know how to script and do the maths on large data. Excel stops working at a million rows. At that point you need a data team," he says.
“While the explosion of data has created new opportunities to compete strategically, there is a widening gap between what can be done and what is being used by companies. This gap comes from the disconnect between data teams and decision-makers. Big data only makes that gap wider, not narrower.
"You close the gap by having data teams develop their listening skills and relationship-building skills. These teams should be completely in sync with decision-makers. With understanding comes trust, and with trust comes action,” says Beckler.
More on data science
- Google’s chief economist examines the data scientist factor
- Buy or build afflicts data scientist capability
- Sherlock Holmes today is a data scientist
- Strata speaker cites key characteristics for success in data science
- Strata spotlights data science in media sector
- Data scientist role requires predictive analysis, expert says
- Data science: Mining available info for hidden value
McKinsey on data science skills gap
Michael Chui, principal at McKinsey, is a co-author of the influential 2011 McKinsey Global Institute report on big data, entitled Big data: The next frontier for innovation, competition, and productivity.
The firm is advising clients how to organise for big data and analytics, and how to find the necessary talent. Data scientists are rare, he says, but they did exist in the world before the term. "They tend to combine knowledge of advanced statistics and machine learning with business savvy.”
He sees the term itself as important to the extent that it reflects an understanding that effective data use can be a basis of competition.
McKinsey, in its 2011 report, pointed out the looming gap in supply of both data science and data-savvy business manager talent. It has spoken to educational institutions about filling the gap.
“Online companies, some marketing departments – in retail, for example – and financial institutions can be sources of talent,” the report said.
IBM’s data science cadre
IBM's Matin Jouzdani says the business analytics practice at the firm, which has about 50 consultants in the UK, has been in place for some 28 years. When recruiting, they look for mathematical aptitude, but also business acumen and customer focus. They get candidates to solve a business problem that involves some number crunching, and they further assess problem-solving skills in a group environment.
When recruiting straight from college, the firm is looking for potential to be comfortable with a variety of tools and techniques. About two-thirds of its recruits come this way.
But it is also important that consultants are comfortable in front of a client, explaining complex analytics clearly, he says.
The people who understand the data in your business are already there. Give them the tools
Matin Jouzdani, IBM
Jouzdani sees value in the way that the big data community is forcing a refocus on the business value of data, and describes the term data scientist as “clever" and "catchy”. Moreover, because some big data technologies are not yet mature, he says there is a chance for smart analysts to get a jump on the market.
For large organisations, however, it is not so much a question of starting afresh with a new data scientist capability as building incrementally on what they have already got, he says.
As for the democratisation of data, Jouzdani says big data will force organisations to collaborate cross-functionally at a lower level than in the past. And there is a consensus on the importance of data.
“To do data analysis well is difficult. That is why the people who are able to do great things with data are being called scientists. I’d like to challenge that. The people who understand the data in your business are already there. Give them the tools,” he says.
Masters in data science
Mark Whitehorn leads a new data science Masters programme, alongside one in business intelligence (BI), at the University of Dundee.
This is the first year of the data science course, which has 17 students, mostly part-time. Most do not have a formal background in mathematics. “If you are motivated enough, and intelligent, then statistics is not insuperably difficult,” he says. "As important is the ability to communicate with users.”
Businesses need people who really understand their own data
Mark Whitehorn, University of Dundee
Simple curiosity is also very important, according to Whitehorn. “A good test is: have you sat down to play with data at 7pm, and suddenly realised it is 2am and you’ve not noticed the room is freezing?”
Not all data scientists have to be Bletchley Park level, but they do tend to come from the top few percent of the population in terms of intelligence, he says.
“Women also have an advantage in this field, since they tend to be better communicators in general. And they are under-recruited in this area,” he adds.
Whitehorn’s advice for organisations that are thinking of building a data science capability is to buy it in from suppliers initially. “But if you have a longer-term need for big data, then build your own capability," he says. "Businesses need people who really understand their own data.”
Whitehorn also says that existing BI staff can become data scientists. “They are BI architects, not, as some would say, report monkeys,” he says.
"This is fun," he concludes. "Don’t go into it thinking you will make money. If data has always intrigued you, then rejoice in the fact that, finally, the world has recognised that data is important.”