Learning certifiably optimal rule lists for categorical data
From MaRDI portal
Abstract: We present the design and implementation of a custom discrete optimization technique for building rule lists over a categorical feature space. Our algorithm produces rule lists with optimal training performance, according to the regularized empirical risk, with a certificate of optimality. By leveraging algorithmic bounds, efficient data structures, and computational reuse, we achieve several orders of magnitude speedup in time and a massive reduction of memory consumption. We demonstrate that our approach produces optimal rule lists on practical problems in seconds. Our results indicate that it is possible to construct optimal sparse rule lists that are approximately as accurate as the COMPAS proprietary risk prediction tool on data from Broward County, Florida, but that are completely interpretable. This framework is a novel alternative to CART and other decision tree methods for interpretable modeling.
Recommendations
- Leveraging integer linear programming to learn optimal fair rule lists
- Learning customized and optimized lists of rules with mathematical programming
- A Bayesian framework for learning rule sets for interpretable classification
- Optimal decision trees for categorical data via integer programming
- Learning Optimal Decision Sets and Lists with SAT
Cites work
- scientific article; zbMATH DE number 3860199 (Why is no real title available?)
- scientific article; zbMATH DE number 1301669 (Why is no real title available?)
- A Bayesian CART algorithm
- A Bayesian framework for learning rule sets for interpretable classification
- A Computational Study of Search Strategies for Mixed Integer Programming
- An experimental evaluation of simplicity in rule learning
- BART: Bayesian additive regression trees
- Bayesian treed models
- Finding a short and accurate decision rule in disjunctive normal form by exhaustive search
- Interpretable classifiers using rules and Bayesian analysis: building a better stroke prediction model
- Learning theory analysis for association rules and sequential event prediction
- Learning with decision lists of data-dependent features
- Supersparse linear integer models for optimized medical scoring systems
- To explain or to predict?
- Using decision lists to construct interpretable and parsimonious treatment regimes
- Very simple classification rules perform well on most commonly used datasets
Cited in
(24)- Efficient Learning of Interpretable Classification Rules
- Learning Optimal Decision Sets and Lists with SAT
- A Bayesian framework for learning rule sets for interpretable classification
- SAT-based rigorous explanations for decision lists
- Causal Rule Sets for Identifying Subgroups with Enhanced Treatment Effects
- A confidence machine for sparse high-order interaction model
- Improving fairness generalization through a sample-robust optimization method
- Learning optimized risk scores
- Interpretable machine learning: fundamental principles and 10 grand challenges
- Synergies between machine learning and reasoning -- an introduction by the Kay R. Amel group
- Learning customized and optimized lists of rules with mathematical programming
- AI 2005: Advances in Artificial Intelligence
- The explanation game: a formal framework for interpretable machine learning
- Classification rules in relaxed logical form
- Logic explained networks
- Considerations when learning additive explanations for black-box models
- Visualizing the Implicit Model Selection Tradeoff
- SIRUS: stable and interpretable RUle set for classification
- scientific article; zbMATH DE number 7415080 (Why is no real title available?)
- corels
- Robust subgroup discovery. Discovering subgroup lists using MDL
- SAT-based optimal classification trees for non-binary data
- scientific article; zbMATH DE number 7625179 (Why is no real title available?)
- Leveraging integer linear programming to learn optimal fair rule lists
This page was built for publication: Learning certifiably optimal rule lists for categorical data
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q115375)
