A Family of Metrics for Clustering Algorithms
This work addresses the need for standardized evaluation in clustering for researchers, but it appears incremental as it builds on existing concepts without introducing a new paradigm.
The paper tackles the problem of evaluating clustering algorithms by proposing a family of metrics that score features like stability and noise sensitivity, resulting in a sample set of scores derived from empirical computations.
We give the motivation for scoring clustering algorithms and a metric $M : A \rightarrow \mathbb{N}$ from the set of clustering algorithms to the natural numbers which we realize as \begin{equation} M(A) = \sum_i α_i |f_i - β_i|^{w_i} \end{equation} where $α_i,β_i,w_i$ are parameters used for scoring the feature $f_i$, which is computed empirically.. We give a method by which one can score features such as stability, noise sensitivity, etc and derive the necessary parameters. We conclude by giving a sample set of scores.