Hackathon · MHCLG · GDS · i.AI (Leeds) · Apr 30 2025
Forecasting housing delivery at the Local Government Innovation Hackathon
In spring 2025 I spent two days in Leeds at the Local Government Innovation Hackathon: 150 planners, technologists and civil servants, run by MHCLG together with the Government Digital Service and the Incubator for AI. There were four real-world problem statements on the table of which my team selected predicting current and future housing demand.
We took that and ran at it from the supply-and-affordability side: could we forecast future housing delivery and affordability at local authority level, and maybe say something about the determinants?
My bit
This was a team effort with people I met at the hackathon including local authority planners, NHS data scientists and UX designers. I spent my time on the core data and modelling of the forecast: pulling together the local-authority-level inputs, getting a forecasting approach working in less than a day, and trying to make the determinants interpretable using a pretty standard random forest + SHAP approach. Our presentation slides can be found here.
How it went
Out of 25 teams we came away with an honorary mention, finishing 4th, which we were pretty pleased with given some of the talent in the room. Would highly recommend attending such hackathons for the interested, if only to bump into some friendly faces from i.AI.