Supercharged: Inside the ONS plan to become a data-science 'powerhouse'
Five years after being established, the Data Science Campus of the ONS wants to do more to help address government's biggest policy issues – while still retaining its innovative edge. PublicTechnology finds out more.
Over the last decade, many advocates of digital transformation – both within and without Whitehall – have suggested that, for all the differences between the two, government could learn something from start-up culture. The ethos of the tech disruptor has often been invoked in relation to the work of the Government Digital Service, particularly in its early years.
A national statistical office with a history dating back to the Second World War does not share a lot of obvious common ground with a trendy young app publisher from Hoxton or Silicon Valley.
But, in 2017, the Office for National Statistics launched an entity that could be considered its own start-up.
The Data Science Campus, based at the ONS’s Newport headquarters, was created to provide an opportunity to innovate – including the time and space to experiment with no guarantee of achieving the desired outcome.
Since then, the facility has delivered 87 projects in which it has sought to gain unexpected insights from unexplored data sources. This has included studying the likes of quarterly VAT returns and the movement of ships in and out of the UK ports to provide indicators of the country’s economic performance, and using artificial intelligence-powered natural language processing techniques to help Brexit negotiators comb through responses to public consultations on trade tariffs.
The campus also has a remit to serve as a hub for data-science professionals across the public sector, as well as to increase expertise, both among specialists and the wider workforce – including 1,000 managers that have gone through a Data Masterclass for Senior Leaders.
All of which means that, after five years – and having grown from eight employees to more than 80 – the campus is ready to move into the next stage of its development.
“We are making this transition from start-up to powerhouse,” Arthur Turrell, recently appointed as acting director, tells PublicTechnology.
“I think permanent secretaries get that data science is the key to delivering lots of benefits, but also to deliver more and better analytical insights. Because what policymakers really want to know is why something happens.”
Arthur Turrell, ONS
He arrived at the facility – which also operates from locations in London, Titchfield, and Darlington – after five years as a senior research economist at the Bank of England, which Turrell says “got into data science quite early”, allowing him to explore emerging analytical techniques.
His time at the central bank included a six-month secondment at the US Federal Reserve Board in Washington DC.
Earlier in his career, Turrell spent a number of years in academia and, last summer, he also published a popular science book: The Star Builders: Nuclear Fusion and the Race to Power the Planet.
He joined the campus in 2021 as deputy director and has now taken acting charge of the organisaiton following the departure of former chief Tom Smith.
He says he was attracted by “the chance to apply data science to some of the most important policy questions” currently facing public service. He identifies three core “USPs” of the campus.
"One is access to the best data in the country,” he adds. “Two, we have the wider public sector remit – we are not just about central government, we are about devolved administrations, ALBs [and others]; we care about delivering projects with them and building data-science capability. Three, is our appetite for innovation and research. We have to strike the balance between working with policy-relevant things and working [on longer-term] projects. We still want to be a hub for data science: we need to recognise that there are now so many data scientists in the rest of government, that it is also about being a hub for them, where they can come and collaborate.”
‘The leading edge’
As Turrell alludes to, the intended evolution into a “powerhouse” requires the campus to balance two roles, acting not only as an innovator – with an in-built tolerance for failure that is far higher than would be accepted elsewhere in government – but also playing an ever-greater role in supporting major projects and policy imperatives.
Turrell stresses the importance of maintaining an appetite for experimentation – which he says is symbiotic to the unit’s efforts in helping to deliver urgent programmes of work.
“That we innovate and do research – that is one of our USPs. If we are going to be that guiding light, we need to be at the leading edge, and innovation and research is part of that,” he says. “I think it is increasingly important over time – especially when we get bigger in scale – that part of the vision for the campus is, as well as doing the short-term, high-priority policy-need stuff, we keep space for innovation and research. And that really benefits the short-term work as well: when there is a crisis and we need to work fast and respond to the policy demands of the day, that innovation helps.”
Of course, government’s work over the past two years has largely consisted of responding to crisis, in the shape of a global pandemic. The Data Science Campus has played its part in helping shape this response with initiatives including the analysis of data from CCTV cameras to help better understand social distancing in urban areas, its relationship with the spread of coronavirus, as well as the impact on city-centre commerce as restrictions were lifted.
Turrell points out that the cameras are typically owned by local authorities, with imagery often already made available publicly for use by transport sector professionals.
The campus studied new and historical data – beginning from March 2020 – drawn from cameras in seven parts of the UK: Durham; London; Southend; Reading; Manchester; North East England; and Northern Ireland. The locations were chosen to provide a range of settings, from rural areas, through small and mid-sized towns and onto big cities. A cloud computing environment was created, with automated feeds providing updated information every 10 minutes.
A report on the project published by ONS concludes that “traffic camera data can offer tremendous value to public authorities by providing real-time statistics to monitor the busyness of local populations… [which] can inform policy interventions such as those seen during the coronavirus pandemic”.
The campus hopes its methods can once again play a supporting role in enabling government’s response to other looming emergencies.
“I want us to make a major contribution to the cost-of-living crisis and levelling up,” Turrell says.
Number of government senior leaders that have taken the campus’s data masterclass
Intervals at which the campus gathered CCTV images from thousands of cameras around the country to help understand urban movement trends during the pandemic
Approximate number of staff at the campus, having risen from just eight five years ago
27 March 2017
Date on which the campus was founded
Number of projects data scientists have completed, with 31 currently ongoing
The ONS has previously experimented with the use of so-called web scraping technologies to track online retail prices, and the campus intends to further this work in the coming months by keeping ongoing tabs on supermarket pricing, as listed in their online stores.
Data scientists will also look to build on work undertaken during the pandemic to provide insights into global supply chains. As part of the campus’s economic indicators work, worldwide GPS information is already used to track the movement of ships.
“We knew that the ships that were bringing goods into the country were backed up outside ports,” Turrell says.
The ambition is now to derive information from bills of lading – the documents issued by shipping firms to importers and containing information on the cargo being carried.
Turrell says that this could enable better understanding of levels of supply of goods, and help spot patterns and possible shortages.
“We know where the ships are, but what we would like to know is what is on them… and what does it mean if a ship is, for example, 70% trousers and 30% iPhones? And how many units does this equate to?,” he adds.
Such a project – involving the collection and collation of big data sources – is an archetype of the kind of work in which the campus specialises.
Big data refers to large volumes of information that is, invariably, complex and unstructured – making it much harder to organise and analyse. This contrasts somewhat with the traditional work of the ONS, in which information is precisely gathered, often through surveys designed especially for that purpose.
Turrell believes that the tried-and-trusted methods of statistical research can combine with the more experimental techniques of the campus.
“I think there is a complementary role,” he says. “If there is a survey on something, we can work out where it is weakest and strongest, or if there are efficiencies to be made – because surveys are expensive.”
“We have the most exciting data in the UK – whether it be economic or social. If you are in a private firm, the main data that you have is about your customers, or sales or products. What data scientists really want to do is data science – and the campus a fantastic place to come and do that.”
Arthur Turrell, ONS
Innovative data science has begun to play a part not just in the rest of the statistical profession, but in the work of government more widely; last year, the campus delivered training to 4,000 people, encompassing “everyone from five-year-olds to permanent secretaries”, according to the campus director.
PublicTechnology wonders whether Turrell thinks the latter group have grasped the potential of data science, and are supportive of its use by policy and delivery professionals.
“My sense is a strong ‘yes’,” he says. “I think they get that data science is the key to delivering lots of benefits, but also to deliver more and better analytical insights. Because what policymakers really want to know is why something happens.”
To better enable this to happen, the campus runs two-month accelerator programmes for government analysts to study data science and data visualisation.
“We pair up someone with a mentor; they can learn some data science from us and take it back to their home department,” Turrell says.
He adds that he considers it “a great success of data science if people are using” its methods without necessarily knowing that they are doing so.
“Not all data science is going to be embedded… but there are some things which are just so useful and, hopefully, a bit more accessible, and my ambition is that these things will be completely embedded,” he says.
The campus also wishes to remain as a central hub for experts – and attract and retain some of the best talent that the sector has to offer.
Government has long faced difficulties in competing for staff with potential private sector employers. Such challenges are especially steep in a field where skills are so in demand.
“I may be biased, but I think we have a really, really strong proposition in many ways,” Turrell says. “Firstly, we have the most exciting data in the UK – whether it be economic or social. If you are in a private firm, the main data that you have is about your customers, or sales or products.”
He adds. “What data scientists really want to do is data science – and the campus a fantastic place to come and do that. Public policy is really interesting; if people want to make a difference and make an impact on the world, then you can do that – you can work on the UK’s most important issues, whether it is supply chains, the cost-of-living crisis, Covid-19: we are working on all of those things, and this is a place where you can really make a difference. There are obviously always constraints in the public sector in terms of pay – and that is challenging, in a world where data scientists are so well remunerated. But we have an appetite for innovation and research, and you can really make a difference.”
As part of its journey from start-up to powerhouse, the campus is “really thinking about our delivery model”, according to the director. This model is summarised as: explore; embed; enable.
To support this, the position of head of delivery was recently created. Sharon Hill, a 30-year veteran of the ONS – and an expert in agile project methodologies – has been appointed to the grade six role. The post sits at the head of a delivery-management team, which will be “borrowing things from the software-developing world” and applying them to the data science projects, Turrell (picture right) says.
The likely focus of these projects is reflected by the area of interest of each of the nine data science “squads” housed within the campus.
These include: levelling up; economic insights; trade, ships and global supply chain; data capability, mobility; Covid-19; and net zero.
The penultimate squad is run jointly with the Foreign, Commonwealth and Development Office and focuses on international development.
The last of the units is dedicated to synthetic data – which uses real information to automatically generate comparable anonymised data sets. This allows analysts to study individual, rather than aggregate trends – without compromising privacy.
“Synthetic data is an emerging area of technology; if we could unlock it, there could be big benefits,” Turrell says.
Five years after starting up, this is one of a growing number of examples of a desire to think big.
Parliamentary committee writes to department urging greater openness
Government operations leader wants departments to make better use of the ‘huge amounts of data’ at their disposal
Workers delivering webchat and telephone service via outsourced deal vote for six-day walkout
Contract with specialist services firm will enable department to convert graduate contractors into full-time civil servants