Supporting troubled families. Ensuring children are school-ready by the age of five. Tackling diabetes. Predicting and preventing fires. Identifying rogue landlords. Combating alcohol abuse.
These are just a few of the things that UK cities are already tackling with data science. On 12 September, practitioners from Manchester, Bath, London, Essex, Sunderland, Birmingham, Sheffield and Cardiff gathered at Nesta to share their experiences and discuss how to overcome the hurdles that stand in the way of joining up, analysing and acting upon data at a city scale.
Sarah Henry, Head of Intelligence and Performance at Manchester City Council, spoke about GM-Connect (an initiative inspired by the Mayor’s Office of Data Analytics in New York City), whose aim is to put in place the data infrastructure and capabilities to make a success of city devolution. With a long list of projects in the pipeline, an initial pilot has focused on creating a “child passport”: federating intelligence so that all agencies have a single view of what is known about vulnerable children.
A specific part of this work entails sharing data on school absences between the region’s ten local authorities. 58% of Greater Manchester’s population live less than two miles from a local authority boundary. Consequently, many children are educated in a neighbouring council area. Given that school absence is a key factor in identifying families in need of support, GM-Connect will help ensure that the arbitrary silos of local government boundaries don’t stand in the way of protecting some of the region’s most vulnerable families.
Richard Puleston and Will Bibby presented on Essex County Council’s work to help identify children at risk of not being school-ready by the age of five. Starting with a pilot in Vange – a ward where 39% of children are estimated to be living in poverty – they are bringing together and matching pseudonymised housing and benefits data, health visitors’ records, social care, youth offending, drug and alcohol abuse data, and police datasets to predict risk at household level. Partnering with data scientists at the University of Essex, their approach has been to identify children in the ward who were aged five in 2015, analysing their data over the past four years to spot predictive indicators of school readiness (or lack thereof). The ambition is that this work will enable early interventions to be put in place in the second half of 2017.
Jon Poole, Business Intelligence Manager at Bath and North East Somerset Council, spoke about their collaboration with the Institute for Policy Research at the University of Bath and work with the community through the Bath:Hacked project. Among other initiatives, early work has focused on diabetes, blending data from multiple sources to remodel the commissioning of services and better support individuals who could benefit from early lifestyle interventions.
Another strand of work has sought to identify people at risk of financial hardship, using DWP and council tax data. Jon highlighted that much of this work was not about adopting advanced new technologies but rather digitising records in the first place and getting them into a simple spreadsheet. He also emphasised the need for data-driven projects to focus on the right level of detail and in answering policy questions. Trying to investigate issues in too much depth risks hindering action. He outlined how a partnership approach to data and analytics can offer a different way of thinking about some of our challenges and unlock access to new data, such as the Strava Metro analysis.
Andy Mobbs, Business Intelligence Manager at the London Fire Brigade (LFB), made a powerful case that some people are dying in household fires when smarter use of data could have enabled preventative and potentially life-saving measures. He explained that simply mapping the number of fires across London is of limited effectiveness. After all, fires are started by people not places.
His focus has therefore been on looking at the social characteristics leading to fires. By examining past cases of fires and connecting the addresses to Mosaic data (which show demographic information), he has been able to confirm that older people and those on lower incomes are at higher risk – helping create a map of priority postcodes as shown below. This group make up 19% of all London households but experience 31% of all fires in the home and 33% of all fire-related casualties. Working with front-line colleagues, LFB has built up a detailed profile of most likely to die in a fire in their home:
“This is an older person aged over 65 who lives alone. Where they live, or the type of property, doesn’t matter. They will have reduced mobility and find it hard to walk unaided. As such, they will spend most of their time in one room of their home and often this can become a bed/sitting room. This person is a smoker, uses candles or has other naked flames in their home.
“This person also has other health issues. They may have an impaired judgement or become forgetful or disorientated; either through a health issue, or as a result of their medication or from drinking. There may be signs of previous fire ‘near misses’; this could be cigarette burn marks on clothing or furnishing, or scorch marks from cooking or using candles. This person either receives, or would benefit from, some care support (from a relative, neighbour, or care provider).”
Andy’s key point was this: somebody knows these people.
His aim is to use a wide range of data to identify individuals likely to match the description. Key datasets would include the list of households requiring assisted bin collections – the assumption being that if a person has trouble putting the bins out they are also likely to experience other mobility issues. Households registered for the council tax single occupier discount would also help highlight people living alone.
Andrew Collinge, Assistant Director for Intelligence at the GLA spoke about his work to pilot a London Office of Data Analytics. In collaboration with Nesta and the ASI – a data science SME – he and his team are working to tackle the problem of unlicensed HMOs (houses of multiple occupancy). Inspired by the work of two London boroughs, the project is taking on two strands. The first is identifying and then correctly weighting the importance of data fields that correlate with high probability of a property being an HMO. This could include everything from the height of a building (HMOs tend to be over three or more floors) and whether they are above a commercial premise. The second is to look at where rogue landlords known to one borough are also operating properties in others. Andrew shared his view that a greater number of issues could benefit from a more strategic and coordinated approach across the whole of London.
Finally, Liz St Louis, Head of Customer Service Development at Sunderland City Council, spoke about the creation of their in-house intelligence service. Liz emphasised that the council’s data efforts were intended not as a bolt-on but as a core part of everything the organisation does. She gave several examples of their work, one of which was Adult 360, a project to bring together information about a person and their life from across a number of source systems including Social Care, CES, telecare, intermediate care, city hospitals and the police. Since its launch in August, it has helped deliver better and more coordinated care, equipping over 350 health and social care practitioners with a more complete view of all that individual’s interactions – as shown in the mocked up version below.
It’s still early days for data analytics in UK cities. In future blogs I’ll dive into the hurdles these cities reported and the practical steps they are taking to address them. But the positive news is that bit by bit we’re starting to build up more evidence of what actually works and learn where data science really has something to offer public service reform. At a time of spiralling demand and continued pressure on public sector budgets, it’s in all our interests to help them succeed.
Cover Image: HelloRF Zcool / Shutterstock.com
This blog originally appeared at: http://www.nesta.org.uk/blog/rise-and-rise-uk-city-data-analytics