Abstract:Relative Validity Indices (RVIs) such as the Silhouette Width Criterion, Calinski-Harabasz and Davie's Bouldin indices are the most popular tools for evaluating and optimising applications of clustering. Their ability to rank collections of candidate partitions has been used to guide the selection of the number of clusters, and to compare partitions from different clustering algorithms. Beyond these more conventional tasks, many examples can be found in the literature where RVIs have been used to compare and select other aspects of clustering approaches such as data normalisation procedures, data representation methods, and distance measures. The authors are not aware of any studies that have attempted to establish the suitability of RVIs for such comparisons. Moreover, given the impact of these aspects on pairwise similarities, it is not even immediately obvious how RVIs should be implemented when comparing these aspects. In this study, we conducted experiments with seven common RVIs on over 2.7 million clustering partitions for both synthetic and real-world datasets, encompassing feature-vector and time-series data. Our findings suggest that RVIs are not well-suited to these unconventional tasks, and that conclusions drawn from such applications may be misleading. It is recommended that normalisation procedures, representation methods, and distance measures instead be selected using external validation on high quality labelled datasets or carefully designed outcome-oriented objective criteria, both of which should be informed by relevant domain knowledge and clustering aims.

Word Clustering with Validity Indices

On the Index of Cluster Validity

From A-to-Z Review of Clustering Validation Indices

CVAP: Validation for Cluster Analyses

Validity index for clustering with penalizing method

Word Embeddings and Validity Indexes in Fuzzy Clustering

A GROUP OF NEW INDEXES OF CLUSTER VALIDITY

A Validity Index Approach for Network Partitions

A New Validity Index Based on Intra-Cluster Variation and Inter-Cluster Overlap

Performance evaluation of some clustering algorithms and validity indices

Efficient synthetical clustering validity indexes for hierarchical clustering

On the Use of Relative Validity Indices for Comparing Clustering Approaches

Extended multivariate comparison of 68 cluster validity indices. A review

A Fuzzy Cluster Validity Index Based on Clustering Mistake Measures

Comparing clusterings and numbers of clusters by aggregation of calibrated clustering validity indexes

Normalised clustering accuracy: An asymmetric external cluster validity measure

Deep Clustering Evaluation: How to Validate Internal Clustering Validation Measures

A New Validity Index for Fuzzy-Possibilistic C-Means Clustering

Distribution free optimality intervals for clustering

A new validity measure for fuzzy c-means clustering

Fuzzy C-Means Clustering Validity Function Based on Multiple Clustering Performance Evaluation Components