Non?parametric regression for networks

Network data are becoming increasingly available, and so there is a need to develop suitable methodology for statistical analysis. Networks can be represented as graph Laplacian matrices, which type of manifold-valued data. Our main objective estimate regression curve from sample matrices conditional on set Euclidean covariates, example, in dynamic networks where the covariate time. We an adapted Nadaraya–Watson estimator has uniform weak consistency estimation using power metrics. apply Enron email corpus model smooth trends monthly highlight anomalous networks. Another motivating application given linguistics, explores author's writing style over time based word co-occurrence