Professor Wim Hardyns of the Department of Criminology, Criminal and Social Law at Ghent University is developing a pioneering crime prediction and prevention system with the research project Big Data Policing, or the acronym BIGDATPOL. This model uses big data and machine learning to predict and prevent crime at specific locations and times, without infringing on individual privacy. The goal is to have a validated model that can be used by police forces in Belgium and Europe within five years.
Predicting and preventing crime
Crime patterns vary widely by location and time of day, and traditional policing techniques often fall short of effectively predicting and preventing crime. In addition, the use of individually predictive systems is problematic due to ethical concerns and privacy issues
What does BIGDATPOL do?
BIGDATPOL focuses on predicting crime based on time and location, not individuals. By creating a detailed geographic map and dividing it into small 200-by-200-meter boxes, the system collects extensive data from police departments and relevant factors, such as the presence of nightclubs, train stations and weather forecasts. Machine learning techniques are used to identify patterns in the data to make predictions about where and when crime is likely to occur.
What are the benefits?
- Data-driven approach: by using a wide range of data sources and machine learning, BIGDATPOL can make accurate predictions.
- Privacy protection: the system does not target individuals but rather times and locations, thus avoiding privacy issues.
- Scientifically validated: the model is transparent and based on scientifically validated data and methods, unlike some existing systems in other countries.
Future prospects
Initial tests in the Zenne Valley police zone have already shown promise, with accurate predictions and improved crime prevention in high-risk areas. Within five years, the system is expected to be widely implemented in Belgium and possibly other European countries, leading to a decrease in crime, increased security and greater efficiency of police forces.
The predictive model is currently being tested for a year in 19 Flemish police zones. There are also exploratory discussions with Europol and countries such as Serbia, the Netherlands, Hungary and Austria. The ultimate goal is a validated model that will be widely used to predict, deter and detect crime faster, thereby significantly improving the safety of citizens.
Contact information
- Wim Hardyns: researcher profile
- Email: bigdatpol@ugent.be
- Website: https://bigdatpol.com/
- LinkedIn: BIGDATPOL