Dataset of networks used in assessing the Troika algorithm for clique partitioning and community detection

This dataset contains real and randomly generated networks (random graphs) from a study on clique partitioning. In total, there are 145 network files. This includes five categories: 27 ABCD graphs, 20 LFR graphs, 53 real networks, 20 Barabasi-Albert graphs, and 25 portfolio networks (S&P 500) as described in the article linked below this description. The folders describe the five categories of the networks. Each network is provided in .gml format or .pkl format which can be read into a networkX graph object using standard functions from the networkX library in Python. For accessing other networks used in the study, please refer to the article for references to the primary sources of those network data.This dataset is provided under a CC BY-NC-SA Creative Commons v 4.0 license (Attribution-NonCommercial-ShareAlike). This means that other individuals may remix, tweak, and build upon these data non-commercially, as long as:1) They provide citations to this data repository (https://doi.org/10.6084/m9.figshare.28835837) and the following article: Aref S, Ng B (2025) Troika algorithm: Approximate optimization for accurate clique partitioning and clustering of weighted networks. PLOS Complex Syst 2(9): e0000062. https://doi.org/10.1371/journal.pcsy.00000622) They license the new creations under the identical terms.For more information about the data, one may refer to the article below:Aref S, Ng B (2025) Troika algorithm: Approximate optimization for accurate clique partitioning and clustering of weighted networks. PLOS Complex Syst 2(9): e0000062. https://doi.org/10.1371/journal.pcsy.0000062