Abstract
An important problem in network analysis is predicting a node attribute usingboth network covariates, such as graph embedding coordinates or local subgraphcounts, and conventional node covariates, such as demographic characteristics.While standard regression methods that make use of both types of covariates maybe used for prediction, statistical inference is complicated by the fact thatthe nodal summary statistics are often dependent in complex ways. We show thatunder a mild joint exchangeability assumption, a network analog of conformalprediction achieves finite sample validity for a wide range of networkcovariates. We also show that a form of asymptotic conditional validity isachievable. The methods are illustrated on both simulated networks and acitation network dataset.