pyRuleAnalyzer: A python package for enhancing decision rules in tree-based machine learning models
We present pyRuleAnalyzer, a Python package designed to support the refinement of decision rules in machine learning models based on decision trees, such as those implemented with Scikit-learn. The package enables automatic removal of redundant and overly specific rules, as well as manual correction of inaccurate ones. It incorporates evaluation metrics, considering accuracy, recall, precision, and sparsity to assess performance and interpretability before and after refinements. pyRuleAnalyzer supports binary and multiclass classification problems across various domains (e.g., cybersecurity). Once refined, users can deploy the improved model in real-world applications.
Authors
- Lenardo Chaves e Silva (ORCID: https://orcid.org/0000-0002-4509-8270)
- Eduardo Augusto Bezerra (ORCID: https://orcid.org/0000-0002-2191-6064)
- Ângelo Perkusich (ORCID: https://orcid.org/0000-0002-7377-1258)
- Lucas Lopes (ORCID: https://orcid.org/0000-0001-5010-8528)
- Leandro Dias da Silva (ORCID: https://orcid.org/0000-0003-4352-1466)
- Danilo F. S. Santos (ORCID: https://orcid.org/0000-0002-8162-715X)
- Álvaro Sobrinho (ORCID: https://orcid.org/0000-0002-1753-1890)
- Caio Dias Giló
Institutions
- Universidade Federal de Pernambuco (BR)
- Universidade Federal de Campina Grande (BR)
- Universidade Federal Rural de Pernambuco (BR)
- Universidade Federal de Alagoas (BR)
- Universidade Federal Rural do Semi-Árido (BR)
Publication Details
- Journal
- SoftwareX
- Published
- 2025-10-15
- DOI
- https://doi.org/10.1016/j.softx.2025.102404
- Citations
- 1
- Primary Topic
- Explainable Artificial Intelligence (XAI)
- Type
- article
- Field-Weighted Citation Impact
- 2.48