Floriana Gargiulo (Dpt de mathématiques, Université de Namur, Belgique)

Statistical methods in human dynamics

vendredi 21 février 2014, 9h30 - 10h30

Salle de réunion, espace Turing


I will present two different data-driven approaches to study human mobility.
In the first part I will introduce a morphogenetic stochastic model
for generating individual commuting patterns, starting from aggregate
dataset. It is a simple model based on an iterative version of the
gravity laws in transport systems. I show that the model can be
universally calibrated (no parameters are required) and I propose some
tests to prove the robustness of the procedure. This model is
extremely important for the applications in epidemics’ spreading where
commuters’ fluxes are fundamental drifters for the infections.
Secondly I will present a recent work on researchers’ mobility where
statistical tools at different levels (individual trajectories and
aggregated networks) are used to understand the driving forces in the
researchers’ preferential choices for an academic institution. I will
show the strong relevance in the career of the first visited
university (the PhD institution), where a high probability of
returning is observed and that is fundamental for determining the most
probable academic ranks of the universities in the following career
steps.