Mobility and interaction patterns in social networks

The question of analyzing the predictability of human behavior has been widely studied in literature, to unveil how individuals move, how they can be mobilized and, more philosophically, to understand to what extent our decisions are random or whether we are free to choose. As a consequence of humans relate to each other, we also tend to live in groups at different hierarchies in a social way so it is interesting to analyze how individual features and choices affect the global structure of a society. In this work, we explore the limits of human predictability in terms of shopping behavior, observing that, even when we are constrained to a limited set of possible places where we can make a purchase, predicting where the next purchase will happen is not accurately possible to do by only observing the past. The next question is to study how individual decisions affect emergent phenomena such as the economy or information diffusion across a country. We analyze the contents, temporal and mobility patterns extracted from users’ social media publications to build a profile of the geographical regions that allow to predict the unemployment rate. Finally, we also use a mobile phone call dataset to test whether the dynamics at the urban level, how people create and destroy links within a city, affect the inter-urban diffusion of diseases, virus or rumors. Our results suggest that inter-regional structure is robust and does not vary significantly on time so diffusion processes can be well modeled in terms of static properties of the inter-urban network.Programa Oficial de Doctorado en Ingeniería MatemáticaPresidente: Javier Borge Holthoefer.- Secretario: Rubén Cuevas Rumín.- Vocal: Josep Perelló Palo