OptimalTrees Documentation

OptimalTrees contains learners for training optimal decision trees for classification, regression, survival, and prescription problems. This documentation includes:

• quick start guides for each of the problem types that contain a demo of OptimalTrees in action:
  • classification
  • regression
  • survival
  • prescription
  • policy with categorical treatment
  • policy with numeric treatment
  • policy with survival outcome
  • multi-task classification
  • multi-task regression
• details of the various optimal tree learners, including descriptions of the available parameters
• recommended strategies for parameter tuning and selection
• tips and tricks for getting the best results from OptimalTrees
• information on advanced topics such as classification trees with logistic regression in the leaves
• the OptimalTrees API reference

Citing OptimalTrees

If you use Optimal Trees in your work, we kindly ask that you cite the Interpretable AI software modules. We also ask that you reference the original work that first introduced the relevant algorithm:

• Optimal Classification Trees:

  @article{bertsimas2017optimal,
    title={Optimal classification trees},
    author={Bertsimas, Dimitris and Dunn, Jack},
    journal={Machine Learning},
    volume={106},
    number={7},
    pages={1039--1082},
    year={2017},
    publisher={Springer}
  }

• Optimal Regression Trees:

  @book{bertsimas2019machine,
    title={Machine learning under a modern optimization lens},
    author={Bertsimas, Dimitris and Dunn, Jack},
    year={2019},
    publisher={Dynamic Ideas LLC}
  }

• Optimal Survival Trees:

  @article{bertsimas2022optimal,
    title={Optimal survival trees},
    author={Bertsimas, Dimitris and Dunn, Jack and Gibson, Emma and Orfanoudaki, Agni},
    journal={Machine learning},
    volume={111},
    number={8},
    pages={2951--3023},
    year={2022},
    publisher={Springer}
  }

• Optimal Prescriptive Trees:

  @article{bertsimas2019optimal,
    title={Optimal prescriptive trees},
    author={Bertsimas, Dimitris and Dunn, Jack and Mundru, Nishanth},
    journal={INFORMS Journal on Optimization},
    pages={ijoo--2018},
    year={2019},
    publisher={INFORMS}
  }

• Optimal Policy Trees:

  @article{amram2022optimal,
    title={Optimal policy trees},
    author={Amram, Maxime and Dunn, Jack and Zhuo, Ying Daisy},
    journal={Machine Learning},
    volume={111},
    number={7},
    pages={2741--2768},
    year={2022},
    publisher={Springer}
  }