Interpretable Clustering Ensemble

TL;DR

This paper introduces the first interpretable clustering ensemble algorithm that constructs decision trees in the original feature space, balancing performance with transparency for high-stakes applications.

Contribution

It presents a novel interpretable clustering ensemble method using decision trees guided by statistical association tests, filling a gap in the field.

Findings

Achieves comparable performance to state-of-the-art methods.

Maintains interpretability in clustering results.

Provides a new approach for transparent clustering in critical domains.

Abstract

Clustering ensemble has emerged as an important research topic in the field of machine learning. Although numerous methods have been proposed to improve clustering quality, most existing approaches overlook the need for interpretability in high-stakes applications. In domains such as medical diagnosis and financial risk assessment, algorithms must not only be accurate but also interpretable to ensure transparent and trustworthy decision-making. Therefore, to fill the gap of lack of interpretable algorithms in the field of clustering ensemble, we propose the first interpretable clustering ensemble algorithm in the literature. By treating base partitions as categorical variables, our method constructs a decision tree in the original feature space and use the statistical association test to guide the tree building process. Experimental results demonstrate that our algorithm achieves comparable performance to state-of-the-art (SOTA) clustering ensemble methods while maintaining an additional feature of interpretability. To the best of our knowledge, this is the first interpretable algorithm specifically designed for clustering ensemble, offering a new perspective for future research in interpretable clustering.