Individual Human Mobility Models for sustainable cities applications

Humans are inherently mobile creatures. The way we move around ourenvironment has consequences for a wide range of problems, including the designof efficient transportation systems and the planning of urban areas. Having goodprediction models able to abstract and infer human mobility behaviours within acity is of extreme importance to improve the urban life.This thesis proposes to study human behavior and dynamics through acombination of techniques from network science and data mining. In the context ofhuman mobility, we use GPS data from vehicles to define trajectories in order tounderstand the mobility patterns. We based our mobility models on the IndividualMobility Networks, a graph representation of users trips that will be presented andused in this thesis. Our work also aims to represent a step towards a reliableMobility Analysis framework, capable to exploit the richness of thespatio-temporal data nowadays available. The work done is an exploration ofmeaningful open challenges, from an efficient Trajectory Segmentation of lowsampling GPS data to the definition of a stable car crash prediction model.From simulation of electric vehicles to the ethics aspects of mobility data usage wehave today many ways to make our cities more sustainable and smart. Anotherpromising perspective is the use of such extracted knowledge to investigate moreextensive topics such as Geographical Transfer Learning and Explainability.Further experimentation has been performed in order to improve thecharacterization of the individual human movements having a more complete andricher picture of that.