Researchers' pursuit for the better understanding of the dynamics of travel and travel behaviour led to a constant advance in data collection methods. One such data collection method, the travel diary, is a common proxy for travel behaviour and its use has a long history in the transportation research community. These diaries summarize information about when, where, why and how people travel by collecting information about trips, and their destination and purpose, and triplegs, and their travel mode. Whereas collecting travel diaries for short periods of time of one day was commonplace due to the high cost of conducting travel surveys, visionary researchers have tried to better understand whether travel and travel behaviour is stable or if, and how, it changes over time by collecting multiple day travel diaries from the same users. While the initial results of these researchers were promising, the high cost of travel surveys and the fill in burden of the survey participants limited the research contribution to the scientific community. Before identifying travel diary collection methods that can be used for long periods of time, an interesting phenomenon started to occur: a steady decrease in the response rate to travel diaries. This meant that the pursuit of understanding the evolution of travel behaviour over time stayed in the scientific community and did not evolve to be used by policy makers and industrial partners.

However, with the development of technologies that can collect trajectory data that describe how people travel, researchers have investigated ways to complement and replace the traditional travel diary collection methods. While the initial efforts were only partially successful because scientists had to convince people to carry devices that they were not used to, the wide adoption of smartphones opened up the possibility of wide-scale trajectory-based travel diary collection and, potentially, for long periods of time. This thesis contributes among the same direction by proposing MEILI, a travel diary collection system, and describes the trajectory collection outlet (Paper I) and the system architecture (Paper II). Furthermore, the process of transforming a trajectory into travel diaries by using machine learning is thoroughly documented (Papers III and IV), together with a robust and objective methodology for comparing different travel diary collection system (Papers V and VI). MEILI is presented in the context of current state of the art (Paper VIII) and the researchers' common interest (Paper IX), and has been used in various case studies for collecting travel diaries (Papers I, V, VI, VII). Finally, since MEILI has been successfully used for collecting travel diaries for a period of one week, a new method for understanding the stability and variability of travel patterns over time has been proposed (Paper X).