An Iranian refugee in the UK was beaten to death and set on fire in his neighbourhood despite sending in calls for help to the police multiple times before his death. After the tragic incident, the Avon and Somerset police pulled together a predictive model retrospectively to identify high risk vulnerable people – the victim was at the top of that list.
“That was a very powerful moment for the force,” says Sean Price, a former Head of Business Improvement at Avon and Somerset and now Director of Industry Solutions (Europe, the Middle East and Africa), Public Sector and Healthcare at Qlik. “If this tool was integrated into the tasking and workforce practices of the organisation, we could understand how to prevent such incidents”. This is now the case and is helping to reduce harm and save lives across 1.6m people.
GovInsider spoke with Price to find out three ways governments are using predictive data tools to improve the lives of citizens.
Early intervention for youth offenders
Hillingdon London Borough Council is combining data from multiple agencies on a data visualisation platform built by Qlik to steer at-risk youths away from risky behaviour. The Council has launched the AXIS project, which aims to “identify and keep Hillingdon’s young people safe from criminal exploitation using smart, innovative analytics to identify, intervene, prevent and disrupt with targeted resources”, says Price. Early intervention teams can then flag high priority cases to relevant agencies who can provide the support to at-risk youths.
The team behind the AXIS project looks at data from across organisations such as the police, emergency departments at hospitals, schools and youth services to identify at-risk people. They look at two kinds of data. “Hard data” includes the crime rates of the neighbourhood they live in and their reoffending rates, amongst other factors. “Have they now been cleared of their drug and alcohol addiction, how often have they gone missing?” Price said, for instance.
The second kind, “soft data”, includes the language young people are using on social media. For example, “‘waps’ is slang for firearms, and ‘hit the stripes’ is slang for dealing drugs”, Price explains. “When you say ‘he or she is tapped’, it implies that the person has lost their mental ability to function.” Intervention officers know to pay attention to such vocabulary, which may be a sign of being involved in crimes.
Combining datasets across several organisations has proven successful for Hillingdon. This project has identified and helped 175 young people who were at risk of falling into crime. A mother whose 14-year-old son benefited from this project said: “Without AXIS my son would have got into serious trouble. They have helped him and me to understand how he could get hurt and made to do things that could hurt other people. He is not hanging around with bad friends anymore and has got into Prince’s Trust which I am really proud of.”
Flood emergency planning
Canada’s capital Ottawa is using data visualisation to design responses to floods. Their platform combines over two million rows of data on residents’ location, real time weather, water levels in rivers and satellite images of flooded areas to help emergency planning staff “visualise the impact of rising water to their community”, says Price.
Emergency planning teams can use this data to predict when flooding would occur, and which areas need to be evacuated. “Rather than reacting and waiting to take the calls that come in, emergency planning teams are able to proactively move people,” says Price.
This tool also allows more targeted evacuation planning, so only people in specific regions that are predicted to flood will be moved. “Sometimes we just evacuate everybody but that’s at a huge cost, and we don’t necessarily have to do that,” notes Price.
Government agencies across the UK and the US are using data to predict how likely an employee is to resign. They have built a predictive model using data on previous employees who have left the organisation. This model is able to highlight a series of predictors that might make someone more likely to leave their job.
Length of service, for instance, could be a factor. Another predictor could be the type of role an employee is in. “Some roles are high stress, high pressure roles that could lead to significant burnout issues,” says Price. A police officer dealing with a large number of child protection cases might encounter cases of rape and abuse very frequently; and road safety officers would often see collisions and severe injuries on a frequent basis.
Being able to predict whether employees would leave a company is important for management of manpower and training. This is especially crucial in healthcare roles like doctors and nurses, which require a lot of staff training, and emergency services, where there’s a lack of trained staff. Having an early indication that an employee might leave allows organisations to provide more support for them. In cases where resignations can’t be prevented, the agencies are better prepared to manage the fallout.
Governments have a lot to gain from using predictive analytics. With Qlik, they are able to step in early to support youths with a high risk of vulnerability, they could protect citizens and minimise damage when flooding occurs; and they could optimise staff management with HR analytics. Prediction allows prevention, and as Price notes, “there’s nothing more powerful than when you’ve got an opportunity to prevent some of the most serious things happening in society.”