452. Margin Optimal Regression Trees
Invited abstract in session MD-13: Recent advances in optimization problems with cardinality constraints, stream Sparsity guarantee and cardinality-constrained (MI)NLPs.
Monday, 16:30-18:30Room: B100/6009
Authors (first author is the speaker)
| 1. | Ilaria Ciocci
|
| Department of Computer, Control and Management Engineering, Sapienza University of Rome | |
| 2. | Marta Monaci
|
| Institute for System Analysis and Computer Science "Antonio Ruberti" (IASI), National Research Council of Italy | |
| 3. | Laura Palagi
|
| Department of Computer, Control, and Management Engineering A. Ruberti, Sapienza University of Rome |
Abstract
Interpretable machine learning models have gained increasing attention in recent years, as they provide explanatory and transparent insights into their decision-making process. Among these models, decision trees have been widely studied thanks to their intuitive structure and inherent interpretability. Along this research line, we extend to regression tasks the Margin Optimal Classification Trees (MARGOT) approach, introduced by D’Onofrio et al. (C&OR, 2024), which embeds Support Vector Machines along the binary tree structure. This leads to a quadratic mixed-integer formulation for optimal regression trees. We address the sparsity of the proposed model by introducing cardinality constraints on the features, selecting only the most relevant ones in order to enhance interpretability. To evaluate both the predictive and optimization performances of our approach, we conduct computational experiments on benchmark datasets, comparing it with state-of-the-art optimal tree methods.
Keywords
- Nonlinear mixed integer optimization
- Computational mathematical optimization
- Optimization for learning and data analysis
Status: accepted
Back to the list of papers