arbiter
JournalsAccount

Stability Selection and Consensus Clustering in R: The R Package sharp

Barbara Bodinier et al.

Journal of Statistical Software2025https://doi.org/10.18637/jss.v112.i05article
ABDC A
Weight
0.48

Abstract

The R package sharp (Stability-enHanced Approaches using Resampling Procedures) provides an integrated framework for stability-enhanced variable selection, graphical modeling and clustering. In stability selection, a feature selection algorithm is combined with a resampling technique to estimate feature selection probabilities. Features with selection proportions above a threshold are considered stably selected. Similarly, a clustering algorithm is applied on multiple subsamples of items to compute co-membership proportions in consensus clustering. The consensus clusters are obtained by clustering using comembership proportions as a measure of similarity. We calibrate the hyper-parameters of stability selection (or consensus clustering) jointly by maximizing a consensus score calculated under the null hypothesis of equiprobability of selection (or co-membership), which characterizes instability. The package offers flexibility in the modeling, includes diagnostic and visualization tools, and allows for parallelization.

5 citations

Open via your library →

Cite this paper

https://doi.org/https://doi.org/10.18637/jss.v112.i05

Or copy a formatted citation

@article{barbara2025,
  title        = {{Stability Selection and Consensus Clustering in R: The R Package sharp}},
  author       = {Barbara Bodinier et al.},
  journal      = {Journal of Statistical Software},
  year         = {2025},
  doi          = {https://doi.org/https://doi.org/10.18637/jss.v112.i05},
}

Paste directly into BibTeX, Zotero, or your reference manager.

Flag this paper

Stability Selection and Consensus Clustering in R: The R Package sharp

Flags are reviewed by the Arbiter methodology team within 5 business days.

Evidence weight

0.48

Balanced mode · F 0.40 / M 0.15 / V 0.05 / R 0.40

F · citation impact0.41 × 0.4 = 0.16
M · momentum0.63 × 0.15 = 0.09
V · venue signal0.50 × 0.05 = 0.03
R · text relevance †0.50 × 0.4 = 0.20

† Text relevance is estimated at 0.50 on the detail page — for your query’s actual relevance score, open this paper from a search result.