The ‘Using Data Science in Policy’ report evaluates the use of data science and machine learning models to inform practical interventions that governments can put into place in policy. It covers six examples focused in four areas: targeting inspections; helping social work professionals to make better decisions; improving the quality of randomised controlled trials (RCTs); and predicting which traffic collisions are most likely to lead to someone being killed or seriously injured.

In targeting inspections, the study found that by using publicly available data published by the Care Quality Commission (CQC) and other sources, 95% of inadequate GP practices could be identified by inspecting only one in five practices.

According to the report, a similar model which only used the public part of the CQC’s Intelligent Monitoring system, would only pinpoint 30% of inadequate surgeries for the same inspection effort.

The Behavioural Insights Team also worked with a local authority to use natural processing to predict which cases that had been flagged for no further action would return within three months, resulting either in a child protection plan or a child being taken into care.

Using text and structured data analysis, researchers were able to predict, 8.3 times better than chance, which cases were likely to be referred back into the system.

David Halpern, chief executive of the Behavioural Insights Team, said: “Predictive analytics and machine learning are opening new pathways to improve public services through the systematic study of subtle variations, and complex relationships, in public service experiences and outcomes. As the results in this report show, there are many ways in which data science techniques can be used to improve government policy and practice.”