NetMix: An Efficient and Practical Model for Applied Network Analysis

Variational EM estimation of mixed-membership stochastic blockmodel for networks, incorporating node-level predictors of mixed-membership vectors, as well as dyad-level predictors. For networks observed over time, the model defines a hidden Markov process that allows the effects of node-level predictors to evolve in discrete, historical periods. In addition, the package offers a variety of utilities for exploring results of estimation, including tools for conducting posterior predictive checks of goodness-of-fit and several plotting functions. The package implements methods described in Olivella, Pratt and Imai (2019) “Dynamic Stochastic Blockmodel Regression for Social Networks: Application to International Conflicts”, available at http://santiagoolivella.info/wp-content/uploads/2018/07/dSBM_Reg.pdf.