Utilizing volunteered geographical information for the benefit of city planners and urban science: Case of Rotterdam

One of the main challenges of the recently popular data science field is establishing a common ground of understanding between technical methods and domain knowledge. Making smart and effective use of data is just as important for public organizations as it is for private organizations as our cities and their problems get more complex with increasing populations and their demands. Addressing this very issue, this thesis focuses on the connection between data science and urban science, mainly on how freely available social media data with geolocational components which is called volunteered geographic information (VGI) here, can be utilized for the benefit of urban science. For this purpose, 3 popular VGI sources; Foursquare, Instagram and Twitter API’s are inspected and compared for their usefulness for data driven urban research. At the first sections of this thesis, a literature review followed by a discussion part is presented about how the smart and effective use of data is beneficial for cities. Then, an event detection application is conceptualized which is used for deriving data and model requirements. This thesis paper sets itself apart by taking a semi-design-oriented approach with real social media data and testing the usefulness of two modeling styles -simple Markov and mixture models- that does not have prior representation in the literature in conjunction with VGI.Complex Systems Engineering and Management (CoSEM