In English

Using Big Data for Human Mobility Patterns - Examining how Twitter data can be used in the study of human movement across space

Gustavo Stolf Jeuken
Göteborg : Chalmers tekniska högskola, 2017. 66 s. Rapportserie för Avdelningen för fysisk resursteori; FRT 2017:07, 2017.
[Examensarbete på avancerad nivå]

Demands for transportation are growing at a fast pace in countries that are experiencing rapid economic growth and urbanisation, such as China, India, Brazil, and Africa. Understanding the spatial and temporal distribution of people and the activities they participate is essential for urban planning, travel demand forecasting, and infrastructure investment. This thesis explores ways in which Twitter data can be useful to understand some important aspects of human mobility, including total travel distance, patterns of mobility and communities. Raw Twitter data was processed to extract relevant information on space and time dimensions and we compare the results across all studied geographies. This information is also fed into a Continuous Time Random Walk (CTRW) model to estimate the average annual distance travelled by people on the same geographies, and we use travel survey data to validate our results. Origin-Destination Matrices (ODM) are generated and the patterns of mobility are visualised on a map and with Rose Diagrams. Finally we use a community detection algorithm to better understand its dynamics of these networks. The validity of our estimates may critically depend on the mathematical models we selected and careful interpretations of the results. Important future work can include continued refinements of our mathematical models to accurately represent total travel distance, identify biases, and further understand how demographics and characteristics of urban infrastructure affect travel demands and mobility patterns.

Nyckelord: big data, human mobility, twitter, ODM

Publikationen registrerades 2017-06-28. Den ändrades senast 2017-06-28

CPL ID: 250155

Detta är en tjänst från Chalmers bibliotek