| Abstract : | During the last decades, researchers from different scientific fields are more and moreengaged in the collection and analysis of data from a network-centric perspective. Owingto the explosion of network data, tremendous developments have taken place in statisticalinference for social network modelling.This thesis discusses the use of models in order to visualize social network data andto summarize the topological structure -both local and global- of a social network. Wemake use of the Latent Position Model to position the network in the latent "social space"and represent the available data. Then we switch to the Latent Position Cluster Model ofHandcock et al. (2007). The model aims at clustering the actors in the latent space. Weapply the models in two real network datasets representing the co-authorship networksfrom faculty members of the Department of Statistics (AUEB and UNIPI) aiming atrepresenting the structure of research from these departments.
|
|---|
Copyright © 2013 Library AUEB. All rights reserved.