The UK's Pensions Regulator (TPR) has built a machine-learning tool to help it focus on pension schemes most at risk of breaching its guidelines. Peter Jackson, head of data at TPR, outlined on a government blog how the watchdog had worked with data scientists to make better use of the "scheme returns", TPR's main source of pension scheme data. Data scientists from the Government Digital Service (GDS) built the machine-learning model to "predict future pension scheme behaviour", Jackson said. "As we have data about pension schemes going back several years, labelled with whether they complied with the requirement to submit a scheme return, we were able to build a supervised machine-learning model that learns from our existing data and can be used to make predictions about how a pension scheme will behave," he said. The final model now being implemented by TPR splits the UK schemes it oversees into 30 groups.