Mitchell centre seminar series
|Dates:||13 February 2019|
|Times:||16:00 - 17:30|
|What is it:||Seminar|
|Organiser:||School of Social Sciences|
Termeh Shafie, University of Manchester
Statistical Entropy Analysis of Network Data
Statistical entropy analysis is a systematic and general exploratory method that can be used to analyse and test complicated dependence structures in data consisting of repeated observations of variables with a common domain and discrete finite range spaces. Combinations of multivariate entropies can reveal that some variables are redundant due to low variation, that is they are almost constant or uniquely determined by other variables. For each variable that is not uniquely determined by others, it might be of interest to know if it is almost uniquely determined with high prediction power by some combinations of other variables. Such dependence measures, as well as independence measures and conditional independence measures, are provided by specific entropy quantities.
In this presentation, a systematic approach to multivariate entropy analysis of network data is shown, and the roles of different entropy tools are explained and illustrated with empirical examples. These applications also illustrate that important social phenomena and processes often are identified with these tools. Even when the analysis relies on aggregated network data, it is a convenient way to explore fundamental dependence structures that will inspire further investigations of the original data.
Travel and Contact Information
Humanities Bridgeford Street