Algorithmic Profiling

How to help jobseekers at risk of long-term unemployment?  The first step for public employment services is to identify the individuals at risk. Profiling systems classify the jobseekers into risk categories and direct services to the ones most in need of services such as training programs or job search assistance. 

With the availability of rich administrative data, profiling efforts can be supported by machine learning algorithms that analyze large amounts of data systematically and make recommendations to the caseworkers. We develop an algorithmic profiling system with Swiss administrative data. A particular focus is placed on discrimination and fairness. 

The goal of the project is to assess whether algorithmic profiling is a way forward for Switzerland and how it can be used responsibly. 

Project by: John Körtner, Giuliano Bonoli