clustAnalytics: An R Package for Assessing Stability and Significance of Communities in Networks

Abstract

This paper introduces the R package [clustAnalytics](https://CRAN.R-project.org/package=clustAnalytics), which comprises a set of criteria for assessing the significance and stability of communities in networks found by any clustering algorithm. [clustAnalytics](https://CRAN.R-project.org/package=clustAnalytics) works with graphs of class [igraph](https://CRAN.R-project.org/package=igraph) from the R-package [igraph](https://CRAN.R-project.org/package=igraph), extended to handle weighted and/or directed graphs. [clustAnalytics](https://CRAN.R-project.org/package=clustAnalytics) provides a set of community scoring functions, and methods to systematically compare their values to those of a suitable null model, which are of use when testing for cluster significance. It also provides a non parametric bootstrap method combined with similarity metrics derived from information theory and combinatorics, useful when testing for cluster stability, as well as a method to synthetically generate a weighted network with a ground truth community structure based on the preferential attachment model construction, producing networks with communities and scale-free degree distribution.