niarules: advancing interpretable machine learning through numerical association rule mining and 3D coral plot visualizationFister, Iztok (Avtor) Emsenhuber, Gerlinde (Avtor) Plümer, Jan Hendrik (Avtor) Fister, Iztok (Avtor) Holzinger, Andreas (Avtor) association rule mininginterpretable machine learningnumerical association rule miningcoral plot visualizationNumerical association rule mining remains comparatively underexplored in interpretable machine learning, largely due to the challenges of handling continuous variables and the limited availability of effective visualization techniques. We introduce niarules, an open-source R package that provides a complete and extensible pipeline for numerical association rule mining, complemented by advanced post-processing and interactive 3D visualization. The package integrates bio-inspired optimization-based rule mining methods within a modular architecture that encompasses data preprocessing, rule mining, and visualization. A novel radial layout engine, implemented in C++, generates Coral Plots, which depict rules sharing a common consequent as radial trees. This design facilitates intuitive exploration of antecedent specificity, alongside key quality measures such as support, confidence, and lift. By combining methodological innovation with user-friendly visualization, niarules lowers the entry barrier to numerical association rule mining and supports the development of explainable AI systems for numerical datasets.Elsevier B.V.20262026-02-11 14:37:26Članek v reviji97012UDK: 004.8COBISS_ID: 267744259DOI: 10.1016/j.softx.2025.102470ISSN pri članku: 2352-7110sl© 2025 The Author(s).