Point process modeling for directed interaction networks - Statistics > MethodologyReport as inadecuate

Point process modeling for directed interaction networks - Statistics > Methodology - Download this document for free, or read online. Document in PDF available to download.

Abstract: Network data often take the form of repeated interactions between senders andreceivers tabulated over time. A primary question to ask of such data is whichtraits and behaviors are predictive of interaction. To answer this question, amodel is introduced for treating directed interactions as a multivariate pointprocess: a Cox multiplicative intensity model using covariates that depend onthe history of the process. Consistency and asymptotic normality are proved forthe resulting partial-likelihood-based estimators under suitable regularityconditions, and an efficient fitting procedure is described. Multicastinteractions-those involving a single sender but multiple receivers-aretreated explicitly. The resulting inferential framework is then employed tomodel message sending behavior in a corporate e-mail network. The analysisgives a precise quantification of which static shared traits and dynamicnetwork effects are predictive of message recipient selection.

Author: Patrick O. Perry, Patrick J. Wolfe

Source: https://arxiv.org/

Related documents