EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

2983. Feature-Based Interpretable Optimization

Invited abstract in session WB-3: Interpretable Optimization Methods and Applications, stream Data Science Meets Optimization.

Wednesday, 10:30-12:00
Room: 1005 (building: 202)

Authors (first author is the speaker)

1. Sebastian Merten
Business Decisions and Data Science, University of Passau
2. Marc Goerigk
Business Decisions and Data Science, University of Passau
3. Michael Hartisch
Data Science, Friedrich-Alexander-Universität Erlangen-Nürnberg
4. Kartikey Sharma
AISST, Zuse Institute Berlin

Abstract

For optimization models to be used in practice, it's crucial that users trust the results. This is influenced among other factors by the interpretability of the solution process. A previously introduced framework for inherently interpretable optimization models proposes to use decision trees to map instances to solutions of the underlying optimization model. Based on this work, we investigate how we can use more general decision rules to further increase interpretability and at the same time give more freedom to the decision maker. These decision rules do not map to a concrete solution, but to a solution set instead, which is characterized by certain features. A framework for MIP formulations as well as heuristics for generating such decision rules are presented. We outline the challenges and opportunities that these methods can present. In particular, we show the gain in solution quality that our approach can provide by using the conventional framework for interpretable optimization models as a benchmark and discuss the relationship between interpretability and performance. All methods are evaluated by using both artificial and real-world data.

Keywords

Status: accepted

Back to the list of papers