This approach has proven to quickly embed data science skills in teams across government, as graduates from the programme directly apply the skills they’ve learnt to their day jobs.

1) Bringing the whole group together once a week is really beneficial. This interaction provides an important opportunity to learn from each others’ experiences. Weekly sessions facilitate discussion as work develops, and helps to refine skills in presenting on data science; key to influencing decision makers and bringing data science to the attention of colleagues. We also bring a flavour of agile working, encouraging participants to ‘fail fast’ by experimenting with different techniques and not being afraid to change tack if more effective methods are found. We facilitate agile ceremonies such as ‘standups’ where participants share brief updates about what they’re working on.

2) The programme started with a few data scientists in GDS and ONS providing mentoring, and as data science teams were established in other depts we have expanded the range of mentors to a wide range of government departments. As well as being crucial to the success of the Accelerator programme, mentoring provides a career development opportunity through supporting participants from a technical and project management perspective and by allowing mentors to contribute to the success of a cross-government capability building programme.

3) We started out in London and have evolved to expand our offer, with six regional hubs now available to participants: in Bristol, London, Manchester, Newcastle, Newport and Sheffield.

4) An important element of the programme is the peer support provided by participants working together in their regional hubs. The whole cohort comes together at several points throughout the three month programme, including at the graduation event where each participant presents their project to their peers and to stakeholders from their home department. Participants are also involved in the Data Science Community of Interest, through regular in-person meetups and through virtual communication channels.

The Accelerator programme runs for 12 weeks, four times a year and is about to welcome its tenth cohort of participants. So far the Accelerator has been responsible for the training of 78 people across 26 departments and public sector bodies. The Accelerator has also expanded into UK regions, and we now host 6 hubs across the country, in London, Sheffield, Bristol, Newport, Newcastle, and Manchester.

We’ve collaborated with partners from across the UK Government to facilitate the Accelerator programme and its expansion, with mentors coming from a diverse range of departments to assist participants in the delivery of their projects. GDS is also a member of the longstanding Government Data Science Partnership (GDSP), alongside the Office for National Statistics (ONS), and the Government Office for Science (GO Science). Through this partnership it has been possible to expand at pace, and ensure high level stakeholders from the respective partners are coordinated in supporting and advocating for the programme.

The Accelerator programme is designed to accommodate public sector workers from a range of organisations, and is under constant review and iteration through regularly scheduled retrospectives and feedback sessions to ensure the experience is tailored to the needs of the participants. Cohort participants are mentored by experienced data scientists as standard, however further partner collaboration with other bodies, including potentially private sector companies, may take place with respect to data access, code sharing, and further learning, these activities take place on a case by case basis.

The Accelerator has delivered a number of impactful outcomes, including projects that measure and predict UK inflation, showing the effect of economic determinants on fiscal outcomes, and understanding the location and migration of demographic groups within cities. Aside from the impacts of the projects themselves, the Accelerator has acted as a way of successfully introducing the concept of data science to UK government departments and agencies, by enabling officials to deliver projects that really exemplify the potential of using new and innovative approaches to address longstanding business problems. This has been a major contributing factor to the trend of data scientist recruitment within departments and agencies, as well as an increase in the number of interested parties for scoping discrete data science projects to address problems or issues. The long term impact of the Accelerator programme will be understood in due course, however at this stage we’ve seen participants return to their home departments/agencies and have an immediate impact in breaking down many of the barriers to the delivery of data science (such as prohibitive contracts with software providers etc), and transformational ways of working take root.

Finding the right blend of mentors and participants, as well as optioning high quality project proposals, and ensuring each person has a day a week (as a minimum) to focus on project work has been a challenge. To mitigate these issues we have created a list of potential mentors for all of the hubs, including their respective specialty areas and preferences for data science tools/techniques. This ensures that the optimal mentors are chosen for any given project. Projects are quality controlled through a sifting process, examining the approaches, tools, existing skill-level of applicants, and plans for dissemination of learnings within the home departments or agency. Securing the time of both participants and mentors is achieved through the signing of Memorandum of Understanding (MOU) contracts by the department line management, committing them to releasing their member of staff to the Accelerator for the requisite amount of time.

For an innovation such as this to be successful there needs to be a willingness to fully plan and design the processes, an appetite from established practitioners to help and mentor prospective participants, and above all a healthy enthusiasm from the applicants themselves and senior leaders in departments and professions. The quality of the deliverables and outcomes of the Accelerator programme rest largely on the extent that participants apply themselves, and we’ve seen that those who find the time to put into developing their projects and improving their domain knowledge and skill-sets have had the greatest impact, both within their departments, and also in their own careers.

This innovation could be replicated to address similar skill gaps within government. The concept of mentoring for development is nothing new, but by formalising it in this way, framing it around the delivery of a central project that will add value to a department, and scaling it across geographical and departmental boundaries, has led to significant impacts that have grown exponentially.

As many studies suggest there is an issue both in the UK and globally in terms of data science training and access, we have learned through the Accelerator programme that with some of the foundation skills, time and space to learn, and some expert mentoring it is possible to develop these new skills within the existing workforce. Within the programme we have learned that it is vital to ensure that applicants possess a certain level of technical skill in basic analysis, as this has proven to be variable and has a huge impact on the quality of the project being developed, as well as the expedience of the project delivery. We have also learned that projects which require access to specific datasets need to also have assurances and proof of access prior to beginning on the accelerator, data access issues are the biggest blocker to successfully delivering data science projects, so we are diligent in ensuring this angle is sufficiently appraised before proceeding with project work.