Safety
Less criminality, more safety
Public authorities are increasingly adopting a broad approach to maintaining a safe society. At the same time, business parks are increasingly found to facilitate unlawful activities, often with regard to drug or people trafficking or money laundering.
Centerdata was recently commissioned by two provincial governments (of Gelderland and Noord-Brabant) to identify significant patterns in data available about local business parks. The idea is that data science techniques could help detect unlawful activities at business parks at an early stage.
Method
We interpret information regarding for instance outdated buildings, vacant buildings, high location mobility, few energy grid connections, or a suspiciously large number of Chamber of Commerce registrations. Unusually high energy use can be another important indicator. These were some of the findings to emerge from a previous pilot regarding unlawful activities that Centerdata performed for the Ministry of Justice.
One of the methods used by Centerdata in the data research for the provinces of Gelderland and Noord-Brabant is a so-called supervised learning algorithm known as Random Forest. With this method, we compare a large number of different data models to determine which model has the best predictive ability.
Our ultimate goal is to develop a risk assessment tool to enable the detection of suspicious signs from a vast amount of information. A certain combination of factors can sometimes imply a larger chance of criminal activity. A combination of such factors is sometimes referred to as a poisonous cocktail.