Abstract:
|
One challenge in making statistical inference from network data is that the relationship between one pair of nodes is potentially dependent on that of any other pair. In the context of dynamic networks, this suggests that the relationship between a pair of nodes is possibly, but very unlikely to be, influential on the relationship between any other pair of nodes at the next time point. This notion can be described statistically with an autoregressive process with a very large but very sparse autoregression parameter. We discuss shrinkage methods for such a parameter.
|