RuleKit 2: Faster and simpler rule learning

TL;DR

The paper addresses the need for scalable, interpretable rule-based learning across classification, regression, and survival tasks on large datasets. It introduces RuleKit 2, combining new algorithms and optimized implementations with a Python package compatible with scikit-learn and a Streamlit browser GUI, replacing the prior RapidMiner dependency. Empirical results show substantial speedups (4–8x across 120 datasets, up to ~230x on large classification tasks) while preserving or improving predictive performance, with regression benefiting from a mean-based induction algorithm. The work demonstrates broad impact through widespread adoption, applications in medicine and industry, and use in teaching, underscoring RuleKit 2's practicality for knowledge discovery and human-in-the-loop decision making.

Abstract

Rules offer an invaluable combination of predictive and descriptive capabilities. Our package for rule-based data analysis, RuleKit, has proven its effectiveness in classification, regression, and survival problems. Here we present its second version. New algorithms and optimized implementations of those previously included, significantly improved the computational performance of our suite, reducing the analysis time of some data sets by two orders of magnitude. The usability of RuleKit 2 is provided by two new components: Python package and browser application with a graphical user interface. The former complies with scikit-learn, the most popular data mining library for Python, allowing RuleKit 2 to be straightforwardly integrated into existing data analysis pipelines. RuleKit 2 is available at GitHub under GNU AGPL 3 license (https://github.com/adaa-polsl/RuleKit)

Figures (5)

• Figure 1: RuleKit 2 architecture.
• Figure 2: RuleKit performance in three selected large data sets (classification, regression, and survival). The experiments were performed in 10-fold cross validation.
• Figure 3:
• Figure 4:
• Figure 5: