Changelog

colSBM 0.4.4

• Switching to using hclust in colSBM::clusterize_bipartite_networks() to give a more consistent clustering output
• Extracting colSBM::compute_bicl_partition() to compute BIC-L for various objects

colSBM 0.4.3

• Adding fit_opts$kmeans_iter to control the number of iterations for kmeans in spectral clustering
• Adding fit_opts$kmeans_nstart to control the number of random starts for kmeans in spectral_clustering

colSBM 0.4.2

• Reworked clustering to have it de-recursified
• Added a clustering function based on graphon distance
• Added a $cluster vector to clustering output
• Removing extract_best_partition() and extract_clustering_dendrogram()

colSBM 0.4.1

• Reworking plot functions to be more flexible and customizable by not relying on R6 plot methods
• Turning plot.bisbmpop into an heatmap plot with possibility to select criterion to show
• Adding a extract_nodes_groups function to extract the nodes groups from a fitted collection

colSBM 0.4.0

• Adding colBiSBM, supporting bipartite collections
• Changing parallelization support, using a colsbm_lapply generic function that the user can customize with their own backend (parallel, future)
• Adding dorebipartite a new dataset containing a bipartite collection dataset

colSBM 0.3.0

• Improved generic plot() functions

colSBM 0.2.1

• BREAKING CHANGE: emission distribution arguments are now named “distribution” instead of “model”
• Fixed an error in extract_best_partition()
• Fixed a bug when using “spectal_init” as a global_opts arguement

colSBM 0.2.0

• Added sparse matrix with the Matrix package leading to less intensive memory usage
• Added extract_best_partition() function
• Improved documentation, references and vignette

colSBM 0.1.1

colSBM 0.1.0

Breaking changes

• Changed the name of EstimateBMPOP() to clusterize_networks()
• Most internal functions are not exported anymore

Cosmetic changes

• Cleaned documentation and user functions
• Added a NEWS.md file to track changes to the package.