Uncover relationships between existing data
The Municipality of Rotterdam supervises citizens that unrightfully obtain social benefit payments. The visitations are intrusive to the citizens and the number of false positives is too high. Therefore the Municipality of Rotterdam was seeking to increase the hit rate of visitations in order to improve the service experience of citizens.
How PA helped
After an explorative analysis of existing internal and external data sources, such as cadastre and feeling of safety in neighbourhoods, PA applied machine learning (random forests) to predict fraud cases. The model uncovered a number of previously unknown relationships between socio-demographical data and social fraud. With these findings, false positives could be reduced.
Benefit for the clients
The Municipality of Rotterdam increased their understanding of factors driving wellfare fraud, was able to reduce false positives for visitations - improving citizen satisfaction. As a side-effect, they better understood how advanced data analytics could be of practical help in their day-to-day operations.