This paper presents a new method for optimizing decision trees using a Markov Decision Problem (MDP) formulation. The approach allows for the creation of optimal decision trees based on various interpretability-performance trade-offs. It provides the user the flexibility to choose the tree that best suits their needs. The authors show that this method competes well with existing algorithms in terms of accuracy and runtime, while offering a range of trees on the interpretability-performance Pareto front.
Publication date: 25 Sep 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2309.12701
