During the last years, data-driven research has become more and more popular. This was made possible by the recent tendency to make datasets publicly available and to provide application programming interfaces (APIs) in different sectors. In addition, individuals generate more and more personal data that can be collected and analysed, due to their increasing use of smartphones, online social networks and other applications of the Web. In this Thesis, we study different aspects of data-driven research, that is data acquisition, method development and applications, in the context of human mobility. The main contributions of the work are divided into three parts. The first part focuses on the study of existing mobility data generated by mobile phone providers. In the second part, we discuss the important properties of data collection campaigns and present our two successful implementations of mobility data acquisition using smartphone applications. In the third part, finally, we discuss the possibility to extend the network paradigm in order to better describe trajectories of the subject of interest. To do so, we develop methods based on higher-order models of Markov dynamics, and study their properties. A particular attention is given to temporal community detection algorithms, aiming at finding densely connected groups of nodes in temporal networks.
|Date of Award||4 Dec 2015|
|Supervisor||Renaud Lambiotte (Supervisor), Anne Lemaitre (President), Timoteo Carletti (Jury), Jean Charles Delvenne (Jury) & Anastasios Noulas (Jury)|
- Human mobility sensing and analysis
- Temporal networks, complex networks
- Data mining
- Complex systems