The Office for National Statistics (ONS) has launched a new data science campus in Newport to develop world-leading expertise to help policymakers, researchers, and businesses.
It will deliver five research programmes under the themes of urban future, society, sustainability, evolving economy and the UK in a global context. It will also offer funding opportunities for PhD candidates and deliver joint projects with partners in the UK and abroad.
Secretary of State for Wales Alun Cairns and Minister for Skills and Science Julie James attended the launch last Monday (27 March) at the ONS, one of the biggest employers in Wales.
Plans for the campus began following an independent review published in March 2016 that recommended the ONS set up a dedicated data hub that could use the public sector’s trove of administrative data to improve economic statistics.
The review suggested that fully capturing the digital economy in official statistics could increase the growth rate of the UK economy by between one-third and two-thirds of a percent.
Projects include analysing street level data and imagery, to produce a dataset of foliage in all urban areas of the country and assess how it indicates social and economic statuses.
Traditionally such a study would involve public polling or reviewing ordinance survey maps, but data science tools and technology such as Google Street View offer more efficient analysis and more comprehensive results.
They are also looking at local business classifications that analyse the strength of the economy in particular industries, and the impacts of political events on sectors involved in imports and exports.
"Our model is that we work on short projects of an absolute maximum of six months, where we ramp up, finish and then ramp down and publish our learnings," MD Tom Smith told Computerworld UK.
There are now 26 full-time staff members at the centre, which will grow to around 60 by the end of the financial year. Candidates with image and time-series analysis techniques and experience using satellite and mobile phones data, street-level imagery, and Internet of Things sensor data are in particularly high demand. They will mix new skills and old, from software developers to classical physicists.
"I see it as a skillset to understand the digital world," says Smith. "There's a huge amount of interesting tools and we're not wedded to any particular technology."
The custom-built facility is constructed on an additional floor of the existing ONS building, with new solar panels on the roof and a lecture space inside.
The decision to transfer the London operations in 2006 to the Newport headquarters was initially detrimental to ONS output. Close to 90 percent of the London staff chose to leave the organisation rather than relocate to the South Wales site or another office in Tichfield.
Interest in the centre's data analytics apprenticeship that began in December suggests attracting talent to Newport is no longer a problem.
"We had 130 applications for apprenticeships for six places at the time. We took eight on. The first three months is formal training, learning Python and R and things like that. Now they’re starting to work on projects, and they're essentially being seconded around the organisation.
"The second round of apprenticeships has gone out and we've had 150 applications for that, so there's huge demand and huge interest."
In addition to the apprenticeship, the first of its kind in the UK, the centre offers a master's in data analytics in government that will begin in September.
The review also recommended greater collaboration with the ONS, other government agencies, universities and the private sector to develop its capacities. Presentations at the launch showed the type of partnerships expected.
Swedish non-profit Flowminder demonstrated how it links satellite imagery of infrastructure and night light in the sky with information from telecoms providers on mobile phone usage to estimate the populations and economies of specific geographic areas.
Its team has tallied up search sensor data from Afghanistan since the 1979 Soviet invasion to estimate how many people live in the country at local levels. The Afghanistan government did not previously have this data, which made it difficult to invest effectively in schools, hospitals and infrastructure.
The governmental organisation Statistics Netherland showed how it uses 50,000 road sensors across the country to provide minute-by-minute updates on vehicles passing by. As vehicle traffic is related to the economy, the Dutch agency is using the information to model GDP figures in advance of official outputs.
"We would love to be building on that, so we're starting to examine street data, road traffic data from DFT and Highways Agency to see whether we can do the same kind of thing here," says Smith.
"The sorts of things we're doing are not in any way going to replace official ONS outputs, but if they can add to some of the things that have already been done, how brilliant."