British Gas revealed that it will form a data science team this week, according to the UK site, V3. The reason may serve as a strategic case for establishing data science teams going forward: The company says the team will help more business users delve into and use its Hadoop data lake.
The announcement reflects a subtle shift in focus, from hiring a team to make Big Data feasible to using a team approach to democratize Big Data.
"We're setting up a data science team to assist our business users so they can fish in the lake themselves," Phil Crannage, head of applications development at British Gas, told V3.
Data teams are not a new idea, of course. Back in 2013, experts suggested that data teams could be used in companies as a way of addressing the shortage of data scientists. That apparently wasn’t a problem for British Gas, which was an early Hadoop adopter. Last year, the company leveraged Hadoop for its first smart meter pilot.
The company plans to expand that program, and hopes its data can be used to improve both efficiencies and customer service — and here’s the tricky part — without escalating costs. That’s a major driver for giving more business users access to that data, Crannage explained.
- Data services expert to manage the data repositories
- Data engineer
- Data manager who focuses on supporting development access and manipulating data content
- Production development includes the person responsible for prepping the data for production, such as simplifying the end-user tool, setting up new algorithms, using new data attributes, etc. Levy said this person would ensure the discoveries maked delivered value to the business.
- The data scientist or whoever is best at analyzing data with an eye for strategic value.
Keep in mind, that list emerged at a time when we thought the data people would bear sole responsibility for extracting value. What British Gas proposes would shift that to an approach that more closely aligns with internal IT software development, where the business makes requests and IT works with the business users to achieve those goals.
For that, organizations might want to look at the data team composition recommended by Rodrigo Rivera, data entrepreneur and founder of analytics company Emplido. Rivera uses Internet companies with 300 or more employees as his starting point. For that initial data science team, Rivera says you’d need a team of five to eight people, including:
- 1 technical project manager
- 1-2 hardcore data scientists to handle modeling
- 3-5 data engineers who deploy the production code
I can’t help but think that any team that will work closely with the business might also need a business-minded project manager, business analyst or data analyst with business savvy. To see what an established data science team looks like, check out LinkedIn’s data team page, which includes a list of team members, complete with titles and links to LinkedIn resumes (of course). You’ll find the team’s blog and a list of its current projects from this page.
Another good site to check out is Harvard’s Institute for Quantitative Social Science Data Science. You’ll find a list of the complete team and titles, as well as an explanation of the team’s approach to deploying data.
Loraine Lawson is a veteran technology reporter and blogger. She currently writes the Integration blog for IT Business Edge, which covers all aspects of integration technology, including data governance and best practices. She has also covered IT/Business Alignment and IT Security for IT Business Edge. Before becoming a freelance writer, Lawson worked at TechRepublic as a site editor and writer, covering mobile, IT management, IT security and other technology trends. Previously, she was a webmaster at the Kentucky Transportation Cabinet and a newspaper journalist. Follow Lawson at Google+ and on Twitter.