On Tuesday 7 September of the RSS 2021 Conference there was a session on UK data science as a global public good. The session was organised by Emily Poskett of the RSS International Development Section and Louise Nolan of the Society's Data Science Section.
Emily introduced the session by reminding us that opinions about the role of data science in national statistics ran the whole gamut of feeling from those who think it has no place whatever to those who think it is the silver bullet and the answer to all our problems! Naturally, the truth lies between. But as Emily said, if we accept that the role of the National Statistical System is to provide the best possible support to policy makers, and we find that data science can improve that support, we should use it.
The first speaker, Ivan Murenzi, deputy director-general of the Rwanda National Institute of Statistics (NISR), described how Rwanda has become one of the leaders in data science in Africa. This is essentially down to having a vision, and then working hard to achieve it.
Rwanda developed its data revolution policy (DRP) back in 2017, and by 2019 data science had become a pillar of the 3rd National Strategy for the Development of Statistics. This put the ball squarely in the NISR’s court to deliver the DRP, but that can only be achieved through close co-operation across government and beyond. Through strategic partnerships with the UK Office for National Statistics (ONS), data science fellowships with the Overseas Development Institute, the UN Economic Commission for Africa (UNECA) and other agencies, capacity was created within Rwanda to such an extent that the country was selected by UNECA to host the Big Data Regional Hub for Africa. But as Ivan explained, they worked hard first to identify the right people to be trained – those who were eager to learn – then to find the right trainers who used real issues as course material, and throughout there was strong and engaged leadership.
Since then projects have demonstrated the benefit of data science techniques to numerous institutions. Ivan cited the National Bank of Rwanda who, during the Covid-19 epidemic, were able to use data from mobile financial transactions to estimate financial flows.
The second speaker was Tim Harris from the joint ONS/FCDO data science hub, which is itself part of the ONS Data Science Campus. The Data Science Campus was established in 2017 to deliver better data to the UK Government. In 2019, the Department for International Development (DFID), now part of the Foreign, Commonwealth and Development Office (FCDO), set up the data science hub to provide better data on international development issues through specific projects, training and mentoring.
Tim briefly outlined the variety of project with which the hub has been engaged: using AI and natural language processing in relation to violence against women; using AIS (automatic identification system for global shipping – literally billions of data points) to monitor port efficiency in Mombasa; using satellite imagery to measure reforestation in Uganda. But the project which had been the most interesting was counting cattle in South Sudan, a highly important and sensitive piece of data but one which involves very small features (the cattle) in a very large landscape.
The third speaker was Rich Leyshon from the ONS data science campus, who spoke about training they had provided to Caribbean statistics offices in collaboration with Statistics Canada, on Reproductive Analytical Pipelines. This takes, say, repetitive tasks and automates them. He cited the example of the Trinidad and Tobago quarterly trade bulletin, which involved someone taking raw data from Eurotrace (the EU trade data package), putting it through Microsoft Access into an Excel spreadsheet, and then finally producing a Word document. This is now automated, easy to update, and easy to reproduce.
The final speaker was Linet Kwamboka, who works for the Global Partnership for Sustainable Development Data in Nairobi but is herself a ground-breaking analyst promoting open data. Linet spoke about the work she and colleagues had been doing to bring data science into the civil service. One project involved a partnership with the African Institute for Mathematical Sciences (AIMS) where data science was introduced into the syllabus. Then, five countries were chosen (Somalia, Ethiopia, Senegal, Malawi and Ghana) to have an AIMS intern for 4 months to train people in data science and also to demonstrate its utility via practical projects.
Linet remarked that data science was still a new idea within the civil service. While accepting that it can be used to speed up data processing, most people have not yet accepted that it can be used to deepen knowledge of the data.
This was altogether an extremely interesting and informative session.
Author
Phil Crook is meetings secretary of the RSS International Development Section and a consultant development statistician. He previously worked for the Department for International Development as a statistics adviser.