Greater Manchester Office of Data Analytics (GM ODA)
Contact: Phil Swan, Julian Cox
Although at an early stage of development, the Greater Manchester ODA has been on our radar for for some time now. Originating from GM-Connect, the wider data and information programme of the Greater Manchester Combined Authority (GMCA), GM ODA will become a Greater Manchester virtual hub, developing solutions on behalf of all organisations that want to use data to improve how decisions are made and actions prioritised.
Origins and funding
The initial thinking around aims, objectives, governance of an ODA originated at the time of GM-Connect foundation, following a successful bid to the 2016 Transformation Challenge Award.
The ODA was given a further boost through the development, in 2017/18, of the Greater Manchester Digital Strategy and action plan.
The ODA is primarily resourced through the GMCA – with support in particular from the Strategy and Research, and Digital teams – as well as through resource in-kind from Greater Manchester District Councils and the GM Health and Social Care Partnership.
Vision and objectives
GMODA’s vision is to promote a way of working and create resources that bring data and problem-owners together in new ways. Unlike existing, frequently fragmented approaches and teams, GMODA will unite the assets of the city-region and experiment collaboratively to create new insights.
Its objectives are to
- increase the collaboration and networking opportunities in the innovation space,
- to improve data skills and technical capabilities of existing teams, and
- to support a better understanding of the data available (and its possible uses for public good) across and beyond the Greater Manchester region.
The host organisation is the Greater Manchester Combined Authority (GMCA). Senior stakeholders of the GM ODA include the chief information officer and both heads of research from the GMCA. The ODA programme reports into the GM Digital Steering Group, which brings together local political and public services leaders, GM tech companies and key representatives from the GM digital community.
The ODA benefits from access to a range of existing Greater Manchester-wide networks, including public service ICT and information governance communities, and key public service reform programmes.
Through pan Greater Manchester work on information governance, over 3,500 data flows have been enabled or are in the process of being enabled. The strategy for information sharing is being owned by a new pan GM Information Board and a common information governance framework is being established using tools like the Information Sharing Gateway.
Greater Manchester has also reached out to its residents and recently completed an extensive study into attitudes towards data sharing which is of fundamental importance to the city region.
Although there is not a physical MoU in place yet, GM ODA is looking to develop a working partnership with more than 10 local authorities: Stockport, Manchester, Wigan, Bury, Trafford, Tameside, Oldham, Rochdale, Bolton, Salford; as well as with wider public services including the NHS and GM Police.
There are no plans at present for the ODA to adopt a physical location.
Project Support for the ODA sits within the GM Digital Service, with research and analytics professional support provided by the GM Strategy and Research team. Technical support has been provided by the Business Intelligence team in Stockport Council.
In addition to the central team, there is a strong analyst network, sourcing capabilities across the region. The GM Analyst Network includes representatives from universities, charities and public sector organisations. It meets regularly, has a twitter account (@GMAnalysts), shares job opportunities and ideas publicly and showcases ongoing projects and ideas at networking events.
Information Sharing: Information Sharing Gateway
Data Sharing: Have previously engaged with some external agencies to investigate secure sharing via a data sharing platform, for example, Witan. At the moment, no data has been shared in the context of the ODA.
Data Storage: An ODA-wide solution has not yet been developed. There is a data warehouse within Stockport Metropolitan Borough Council; future plans include the delivery of a central data warehouse for safely storing data from partners, structured in a way that will allow matching of disparate datasets using common dimension tables through PostgreSQL.
Languages used: Python, R
Data visualisation tools: Tableau
Other Technology: PostgresSQL, GitHub
Disclaimer: the technologies listed above that are currently available for future projects depend on the capabilities of analytical teams across partners. Their goal is to develop data projects using, as much as possible, open source technologies which will allow them to share the code, using open APIs and common standards.
Besides their current work developing synthetic datasets to enable sharing of datasets that are sufficiently anonymised but still reflect the different regions of Greater Manchester, GMODA is at the design stage of their first pilot on school readiness.
School readiness pilot
Problem: Evidence suggests that whether a child is ready for school or not has a large impact on outcomes later on in school and into adulthood. The proportion of children not ready for school in Greater Manchester (GM) is higher than average and in order to support the GM priority on school readiness, the ODA is developing a model to help better understand school readiness. It is exploring the possibility of building a predictive model to help inform decision making and better target resources in this area.
Solution: To build a predictive model that accurately identifies if a child will or won't be school ready before they enter Reception (aged 4). There could also be a potential for individual level risk scores to inform decisions. It is hoped that this will be used to redirect funding or resources.
How they are going to do this: A smaller pilot has been developed using only Stockport Metropolitan Borough Council (SMBC) data, and by using only a handful of variables, for example birth month, sex, area born etc. The model accuracy was at 80 per cent.. SMBC now have a PHD student coming in to peer review what has been done to date for their own model and are aiming to reach 90 to 95 percent accuracy. The challenge is now to build a Greater Manchester wide synthetic dataset, based on this model that is sufficiently anonymised, but where data adequately represents the different regions of Greater Manchester. SMBC are evaluating whether they can share the code/definitions/structure with others. In the smaller pilot, the datasets used were all owned by SMBC and included education data e.g. end of reception data as well as contextual info from own systems on postcode, gender, date of birth, health and their own social care information. There was not a need for data sharing agreements as all date was owned in-house however it is still pseudonymised. Additional information sharing agreements are required for the development of the project.
Work plans for the future
The next, and possibly most challenging step is to develop the ODA structure itself as well as formalising the Information Governance processes. Leadership sign up is less of an issue, considering the appetite in the combined authority for making a better use of analytics.
An existing data sharing initiative that could be considered for further development as a GM ODA pilot is a ‘signposting’ project. This involves police, health, Stockport Homes (social housing association), and the council and it is built on open source software.
The development was assisted by ‘ThoughtWorks’ - and data is blended in the background to provide the police or other agencies a more complete information set about children at risk. This project is already looking at families at large, and as it already involves data sharing across two or more other agencies, it is recommended as first GM ODA pilot.