A partially separable model for dynamic valued networks

The Exponential-family Random Graph Model (ERGM) is a powerful model to fit networks with complex structures. However, for dynamic valued whose observations are matrices of counts that evolve over time, the development ERGM framework still in its infancy. To facilitate modeling dyad value increment and decrement, Partially Separable Temporal proposed networks. parameter learning algorithms inherit state-of-the-art estimation techniques approximate maximum likelihood, by drawing Markov chain Monte Carlo (MCMC) samples conditioning on network from previous time step. ability interpret dynamics forecast temporal trends demonstrated real data.