Tom Smith, director, Data Science Campus, Office for National Statistics

Looking forward to 2021, what are your expectations for data and analytics within your organisation?

Locally, I’m looking forward to implementing machine learning approaches on ONS statistics production - we’re leading a UN programme on this with statistics offices worldwide - as well as increasing the richness of data sources we use to understand the UK’s economy and society.

Nationally, Census day is 21st March – the largest data collection of the decade and a critical source for UK data, statistics, research and analysis. Globally, the sustainable development goals, and particularly data on the environment, will come under increasing interest as we head towards the COP26 meeting chaired by the UK in Glasgow.

In the analysis industry, I’m looking to an increased focus on inclusion and diversity, including but not limited to better tools and understanding of how we evaluate bias in models and training data, lots of work on implementing data science techniques, including machine learning in production, and a harder look at the energy impact of our work (including model training).

But, as a data scientist, I know it’s hard enough accurately identifying what’s really happening now, let alone trying to predict the future. So, let’s see.

Is data for good part of your personal or business agenda for 2021? If so, what form will it take?

The opportunity to do data for good at scale was one of the reasons I joined the UK’s Civil Service in 2017 after working in start-ups and academia, and one of the reasons that I’m still personally excited by the work. The work could not be more important, ensuring that the UK has the data science capability to tackle the biggest challenges and building partnerships across public sector, industry, academia and civil society to bring in the rich and diverse talent needed.

There has never been a more important time to stretch your skills and experience by working on government-scale challenges. So if you’re out there and thinking about how you can do more on data for good, it is definitely worth looking at the digital, data and analysis roles inside government.

What has been your path to power?

One thread running through my career is data for public good. I was lucky as an early job to join an Oxford University research group doing some of the first work with government administrative data, modelling and estimating poverty and deprivation levels, which evolved into the Index of Multiple Deprivation used to target billions of pounds of spending.

Later, I helped launch a spin-out data, analysis and tech company from the research group, and over a decade we worked with hundreds of public sector organisations across the UK and internationally. Gradually, I became more involved with government work, advising on opening-up and using data, and then applied to ONS to lead the Data Science Campus.

A second thread through my career is exploring the art of the possible. Of course, all organisations are under pressure to deliver immediate results, but it’s critical that we also keep the next things in view, horizon scanning and understanding the next opportunities. My MSc and PhD were in AI and Computational Neuroscience, evolving robots to play football (among other things), and I’ve brought this exploratory focus to the Campus where we have a programme to research the value of new data sources, tools and techniques.

A final thread through my career is collaboration and partnerships. No one organisation or sector has a monopoly on talent and we all need to get smarter at working across organisational and sector boundaries. The Campus has partnerships across government, academia and industry, for example, running a hub in the Foreign Commonwealth and Development Office, and I’m particularly excited by the potential for joint programmes and secondments with industry and academia.

What is the proudest achievement of your career to date?

Through my career, I’ve always looked to build teams and capability in industry and public sector. But I’m most proud of helping to ensure that data science is now a critical part of UK government’s ability to respond to the biggest and most urgent questions. It has been an absolute (and intense) privilege to have the opportunity to build the Data Science Campus group to provide tangible data science and capability support to teams across UK government and internationally, developing the data and data science capability for the Joint Biosecurity Centre to support the UK’s Covid response.

Tell us about a career goal or a purpose for your organisation that you are pursuing?

“Better use of data in government” is a pretty good summary of the Campus programme, where we are delivering data science R&D projects, as well as building government data science capability. We have a target under the national data strategy of putting 500 data scientists into government and are relentlessly pursuing this with a mixture of programmes, including recruitment (graduate and apprenticeship schemes), up-skilling (from MSc to bitesize training courses), peer learning and community (including data science festivals and meet-ups, cross-government mentoring accelerator), leadership masterclasses and much more.

How closely aligned to the business are data and analytics both within your own organisation and at an industry level? What helps to bring the two closer together?

As the UK’s national statistics agency, the Office for National Statistics is built on data and analysis, so it’s fair to say that we’re very closely aligned!

What is your view on how to develop a data culture in an organisation, building out data literacy and creating a data-first mindset?

You need three big things to support data-enabled organisations. First, commitment from the top - leaders who invest in their own understanding of the opportunities and limitations for data, and who invest in the infrastructure and skills required to deliver. As a great example of Civil Service support from the top, the data masterclasses for senior leaders was designed in partnership with the Prime Minister’s office data science group. Second, space for teams and product owners to innovate with data, exploring and evaluating different approaches for improving business processes and outputs. Third, a mechanism for scaling-up successful data innovation, deploying into product and outputs while continuing to evaluate success. And the feature that marks out great data-enabled organisations is how rapidly they go round the “exploration-deployment” loop to get innovations into products.