GREMLINS: Adjusting an extended SBM to Multipartite networks

In Demiperimetre/GREMLIN: Generalized Multipartite Networks

Adjusting an extended SBM to Multipartite networks

Description

Generalized multipartite networks consist in the joint observation of several networks implying some common pre-specified groups of individuals. GREMLIM adjusts an adapted version of the popular stochastic block model to multipartite networks, as described in Bar-hen, Barbillon and Donnet (2020) The GREMLINS package provides the following top-level major functions:

• defineNetwork a function to define carefully a single network.

• rMBM a function to simulate a collection of networks involving common functional groups of entities (with various emission distributions).

• multipartiteBM a function to perform inference (model selection and estimation ) of SBM for a multipartite network.

• multipartiteBMFixedModel a function to estimate the parameters of SBM for a multipartite network for fixed numbers of blocks

Details

We also provide some additional functions useful to analyze the results:

• extractClustersMBM a function to extract the clusters in each functional group

• comparClassif a function to compute the Adjusted Rand Index (ARI) between two classifications

• predictMBM a function to compute the predictions once the model has been fitted

• compLikICL a function to compute the Integrated Likelihood and the ICL criteria for the MBM

Author(s)

Pierre Barbillon, Sophie Donnet

References

Bar-Hen, A. and Barbillon, P. & Donnet S. (2020), "Block models for multipartite networks. Applications in ecology and ethnobiology. Journal of Statistical Modelling (to appear)

Demiperimetre/GREMLIN documentation built on March 14, 2023, 12:55 p.m.