• unsupervised learning
• personal mobility analysis
• visual analytics
• kdd
• data visualisation
• tree based data
• data mining
• human mobility

This thesis proposes a new method to encode personal mobility through tree-shaped structures. In particular, it focuses on a variation of the prefix tree architecture in order to organise the user's movements around a root location, distinguishing between incoming and outcoming paths.
The analysis is supported by an interactive visualisation tool specifically designed and developed by the author, and aims to find patterns in the movements of different users, both directly using the trees and representing them through vectors encoding their features.
The key idea of the work is to focus on the "shape" of the user's mobility rather on the specific locations, opening new possibilities to the analysis, with an eye to privacy preservation, and with the goal to present the content in the most accessible way as possible.