Automated Model Selection for Generalized Linear Models
Machine Learning
2025-12-17 v1 Machine Learning
Optimization and Control
Abstract
In this paper, we show how mixed-integer conic optimization can be used to combine feature subset selection with holistic generalized linear models to fully automate the model selection process. Concretely, we directly optimize for the Akaike and Bayesian information criteria while imposing constraints designed to deal with multicollinearity in the feature selection task. Specifically, we propose a novel pairwise correlation constraint that combines the sign coherence constraint with ideas from classical statistical models like Ridge regression and the OSCAR model.
Keywords
Cite
@article{arxiv.2404.16560,
title = {Automated Model Selection for Generalized Linear Models},
author = {Benjamin Schwendinger and Florian Schwendinger and Laura Vana-Gür},
journal= {arXiv preprint arXiv:2404.16560},
year = {2025}
}