- Daily & Weekly newsletters
- Buy & download The Bulletin
- Comment on our articles
‘Predictive policing’ goal of federal police
The federal police are planning to use databanks and algorithms to embark on a programme of ‘predictive policing’ – using statistics and data from the past to predict where, when and what sort of crimes are likely to be committed in the future.
The Dutch have recently wrapped up predictive policing pilot projects, most notably in Amsterdam. Police plan to roll out the system to the rest of the country.
By 2020, all police databanks in Belgium will be brought together in one system, which could be an optimal time to start testing predictive policing, said spokesperson Guy Theyskens on Radio 1 this morning.
Incorporating the data will provide information on how frequently certain crimes are committed in certain areas but will also show what kind of effect external factors have on crime, explained Theyskens. The weather, for instance, or a holiday weekend.
“The biggest challenge is to discern what data is the most relevant for predictive policing,” said Theyskens. “Which information could we or should we use and which not? And the connections between certain data need to be interpreted.”
Once data shows a pattern, police can respond by adjusting patrol routes, times and frequencies. But Theyskens emphasises that it isn’t foolproof. “It is a system that analyses the probability on the basis of multiple bits of information. That means it is always a hypothesis, not a roadmap.”
Photo: Getty Images