Tom Smith: The Data Science Campus – data science for public good
18 September 2017

The data revolution affects all industries, and government is no different. The Data Science Campus at the UK's Office for National Statistics is part of the UK government’s push to strengthen our expertise in data science. We are exploring how new data sources such as earth observation data, images, social media, and data held by government, can help us better understand the economy, global environment and wider society.

Our work broadly covers two areas: doing data science, and building data science skills. Bringing together analysts, data scientists and technologists from across the UK and the wider international community, the Campus acts as a hub and incubator for analysis and capacity-building, working collaboratively with academics, colleagues across government and the public sector, industry and third sector partners to help us push the boundaries of data science and analytics applications.

Data science projects

We currently have 10 research projects underway under our five core themes – the evolving economy, the UK in a global context, society, urban and rural, and sustainability. The projects underway include:

• Our “Superfast gross domestic product (GDP) growth estimates” project is using early returns data from Value Added Tax (VAT) company returns to identify large changes to GDP growth.
• We have been making great progress on using street-level images to develop statistics on the local environment, using image classifiers and convolutional neural networks.
• Our calorie counting project is exploring the discrepancy between self-reported food consumption (we’re eating less) and weight (we’re getting heavier). We have been running simulations using bio-metric data on calorie consumption obtained from Cambridge University, and comparing against self-reported food consumption. The first results from this work were presented last week at the Royal Statistical Society Conference.
• Our FinBins project on classifying 100,000 finance sector organisations is using random forest and XGBoost machine learning techniques to predict classifications using administrative data collected by public agencies, Office for National Statistics (ONS) survey data held at individual company level, and published data from the Financial Conduct Authority on regulatory permissions. As part of this project, we’ve been road-testing the new ONS data infrastructure, simultaneously running algorithms on 500+ processors.

For more info on our projects see the:

• Campus website
• Data Science Campus Projects page

Building data science skills

Alongside delivering projects, we have a big focus on building data skills across government and wider. Next month sees the start of the new MSc in Data Analytics for Government, a programme designed to equip the next generation of government data scientists and statisticians with an advanced set of skills and abilities. The MSc is provided by University College London, Oxford Brookes University and the University of Southampton.

Working with the Government Statistical Service's (GSS) Learning Academy, we launched the UK’s - and possibly the world’s - first data analytics apprenticeship programme. This was a very exciting step for us, taking 8 out of 140 applications in the first round. The group has now finished their training phase, and are joining teams across ONS this month. The second round of apprentices also start this month, and the success of the programme has seen it extended to the rest of ONS, as well as Welsh Government. Based on this pilot, the GSS are now working with government and industry partners across England to develop a common framework for a degree-level apprenticeship in data science.

Our commitment to developing the next generation of data scientists isn’t limited to government. We are currently partnered with the Alan Turing Institute, the UK's national data science institute, and a range of universities across the UK where we undertake joint research projects, and support the work of Masters and PhD students. Nor does our work stop at the UK’s borders. I’m really excited that this month some of our data scientists are in Rwanda and Ethiopia, working with DFID and the UN Economic Commission for Africa, to help strengthen international data science work.

This has been a bit of a whistle-stop tour of our work at the Campus. But having launched in March 2017, we are only just getting started. We are always looking for new research projects and collaborations that help us deliver public good through data science. So if you are interested in any of the following issues then do get in touch:

• gaining a better insight into the UK’s economy and society.
• answering a question that traditional techniques and data sources have been unable to address.
• exploring data sources such as satellite images, text, Internet of Things, social media, big data, Blockchain.
• exploring the power of machine learning / data science methodologies eg artificial intelligence, clustering, random forests, neural networks, text mining.
• building data science capability in your team through collaboration with members of the data science community.

Tom Smith is managing director at the UK government's Data Science Campus, based at the Office for National Statistics, which explores how new data sources such as earth observation data, images and social mediacan help us better understand the economy and society. A data addict with 20 years’ experience using data and analysis to improve public services, Tom originally trained as a physicist with a PhD in evolving neural networks for robot control. Before joining government, Tom was co-founder and chief executive of OCSI, a research and data 'spin-out' company from the University of Oxford working with 100s of government agencies, including leading the government’s Indices of Deprivation used to allocate more than £1Billion per year. He is vice-chair of the Royal Statistical Society Official Statistics section, previously chair of the Environment Agency Data Advisory Group and member of the Open Data User Group ministerial advisory group.

For more information please see: @DataSciCampus or email datasciencecampus@ons.gov.uk

Any views or opinions presented are solely those of the author and do not necessarily represent those of the MRS Census and Geodemographic Group unless otherwise specifically stated.