On 23 October at 14:15, Erki Saluveer will defend his doctoral thesis, "Methodological frameworks for producing official statistics from mobile positioning data", to obtain the degree of Doctor philosophiae in Geoinformatics.
Supervisors:
Dr. Anto Aasa, University of Tartu, Estonia
Professor Rein Ahas †, University of Tartu, Estonia
Opponent:
Dr. Henrikki Tenkanen, Aalto University, Finland
Summary
A town’s registered population tells us little about how many people spend the working day there or visit during the summer. Yet this information matters when planning public transportation services, developing tourism or preparing for emergencies. Mobile phones offer a way to understand these changing patterns. As people use their phones, mobile networks generate records of when and where the devices connect. This dissertation demonstrates how these records can become reliable official statistics about population, tourism and transport. A network record identifies an approximate location, but does not explain whether someone is at home, commuting to work or visiting. Drawing mainly on Estonian data, the research develops and tests methods for interpreting these records across three dimensions. In the spatial dimension, five methods are developed and compared for moving from discrete antenna locations to more precise, grid-based estimates using auxiliary data such as building and road registers, which substantially improves spatial accuracy. In the temporal dimension, fragmented location events are reconstructed into continuous activity traces, enabling the identification of meaningful locations such as home, work-time, second homes, or other regular locations and their changes in time. These regular places then say something about the person as well: someone with a home in one municipality and a work-time location in another is a commuter, and someone with a second regular home is a seasonal resident. This is the social dimension of the data. Added up by municipality, modelled home locations agree closely with the population register, and linking homes to daytime locations yields commuting flows without a travel survey. Tourism shows what the three dimensions together can do. More than half of all visits to Estonia are same-day trips that accommodation statistics never see, and among overnight visitors from some of Estonia's near neighbours, the mobile phone data also record more nights than hotels report, including stays with friends and relatives or in private rentals. The Bank of Estonia (Eesti Pank) has produced the tourism figures in Estonia's balance of payments every month since 2008, and the same approach has since fed into guidelines at Eurostat and the United Nations. The research also addresses a challenge that prevents many countries from adopting similar approaches: data governance. Working with sensitive telecommunications records requires clear institutional arrangements. This thesis sets out five principles for national statistical offices, providing practical guidance for using mobile data responsibly and sustainably.
