(1) Find a network clustering algorithm. Ask Andrew for one.
(2) Incorporate rho into the MCMC by using the estimate in the Ado paper. Not the MLE, but the “dumb” estimate.
(3) Run clustering algorithm on original Y, newY from model with rho, newY from model without rho. Compare plots too.
Also reconstruct the adjacency matricies like Airoldi did.
(4) How well do these models fit? How can we tell? See 5 and 6.
(5) Posterior Predictive Checks: simulate theta from posterior (i.e. step i in mcmc), then generate new Y using sampling distribution. Calculate T(newY) for each step – this is the null distribution of T. Compare with T(Y).
Ideas for T: density, entry of B matrix, number of clusters, take 1 nodes and determine the proportion of ties to each of the other clusters.
(6) When the time comes, we might include the rho parameter in the mcmc…as a latent trait. Ties are marked by a latent variable either as being forced to be 0 or having a blockmodel edge (prob of ZBZ).
(7) Another idea for model comparison is to look at Bayes factors. See notes from meeting and read Kass & Raftery paper.