Optimal Prescriptive Trees
By Dimitris J. Bertsimas, Jack Dunn, Nishanth Mundru. 2019
109 citations recorded by the source corpus.
Motivated by personalized decision making, given observational data [Formula: see text] involving features [Formula: see text], assigned treatments or prescriptions [Formula: see text], and outcomes [Formula: see text], we propose a tree-based algorithm called optimal prescriptive tree (OPT) that uses either constant or linear models in the leaves of the tree to predict the counterfactuals and assign optimal treatments to new samples. We propose an objective function that balances optimality and accuracy. OPTs are interpretable and highly scalable, accommodate multiple treatments, and provide high-quality prescriptions. We report results involving synthetic and real data that show that OPTs either outperform or are comparable with several state-of-the-art methods. Given their combination of interpretability, scalability, generalizability, and performance, OPTs are an attractive alternative for personalized decision making in a variety of areas, such as online advertising and personalized medicine.
Subfields: Statistics and Probability
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Statistics and Probability
- Ryan McNellis
- Carlos Eduardo Pedreira
- Adam N. Elmachtoub
- Vinícius G. Costa
- Kris Johnson Ferreira
- Ali Aouad
Citations in the corpus
- Decision Trees for Decision-Making under the Predict-then-Optimize Framework (2020, unknown): “We remark that the use of decision trees for decision-making problems has seen increased attention in practice and recent literature due to their interpretability (Kallus 2017, Elmachtoub et al. 2017, Ciocan and Miˇsi´c 2018, Bertsimas et al. 2019, Aghaei et al. 2019, Aouad et al. 2019).”
- Recent advances in decision trees: an updated survey (2022, methodology): “Both algorithms were subsequently built upon: OCT was extended for regression (Dunn 2018) and prescriptive tasks (Bertsimas et al. 2019), while DTIP was extended for fuzzy trees (Rhuggenaath et al. 2018) and its authors proposed a novel binary formulation called BinOCT (Verwer and Zhang 2019).”
- Market Segmentation Trees (2019, methodology): “(2013), and Bertsimas et al. (2021)). We argue that the approach of first clustering and then fitting response models within each cluster suffers from a fundamental limitation – namely, the resulting market segmentation does not take into account the predictive accuracy of the resulting collection o”