FCCP: Fairness by Chance Constraints
Project Leader: Prof. Dr. Jörg Rambau
Contact: Prof. Dr. Jörg Rambau
Project start: 2024/06/01
Project team:
External partners:
- Associate Professor Bismark Singh, University of Southampton, United Kingdom
Support: Humboldt-Centre of the University of Bayreuth
Project description:
Application Background
Many optimization challenges concern the decisions of more than one stakeholder. The system-optimal solution does not by itself balance the benefits between the various actors. Thus, sometimes optimized solutions are not put into practice, since one or more stakeholders see insufficiently high benefits for themselves compared to others. Fairness constraints can be added to any optimization problem. However, sometimes these produce over-conservative solutions in the presence of uncertain data. The idea in this project is to use Chance Constraints to model fairness. Such chance constraints then require that the fairness is in place most of the times. Examples for such problems are the distribution of apparel among many branches (having an interest to perform well on an individual basis. Or the location of recycling centers (each of individual importance, positive and negative, to the people live in the vicinity).
Contribution to the Mission of MODUS
This topic belongs to the large field of Stochastic Optimization. We aim at solution methods from Mixed Integer Linear Programming (MILP).
Results
We have made suggestions on how to inject fairness by chance constraints into the lot-type design problem that was investigated earlier in a cooperation with NKD. The next step is to generate computational results to assess the impact on the solution quality.