View Item 
        •   Utrecht University Student Theses Repository Home
        • UU Theses Repository
        • Theses
        • View Item
        •   Utrecht University Student Theses Repository Home
        • UU Theses Repository
        • Theses
        • View Item
        JavaScript is disabled for your browser. Some features of this site may not work without it.

        Browse

        All of UU Student Theses RepositoryBy Issue DateAuthorsTitlesSubjectsThis CollectionBy Issue DateAuthorsTitlesSubjects

        Predicting City Pass use among low-income citizens of Amsterdam

        Thumbnail
        View/Open
        Master_thesis_sissi_cao_4016351_final_upload_version.pdf (4.483Mb)
        Master_thesis_sissi_cao_4016351_final_upload_version_corrected.pdf (4.483Mb)
        Publication date
        2019
        Author
        Cao, W.
        Metadata
        Show full item record
        Summary
        The City Pass is one of several Poverty Reduction programmes from the Municipality of Amsterdam. It enables low-income citizens in Amsterdam to partake in a wide range of activities, either for free or with a discount. These include cultural and sport locations. This study investigates how well City Pass use can be predicted and understood with machine learning techniques, with a focus on interpretability. Interpretability includes insights such as feature importance from the supervised machine learning models. This can be valuable in creating more understanding of City Pass user behaviour. City Pass use encompasses unique use, as well as cultural and sport participation. Unique use refers to whether an owner of a City Pass actively uses it. Participation refers to having visited a type of location as outcome. Three existing supervised machine learning models and an unsupervised machine learning model were implemented for this task. Data included user level information such as demographic data, as well as neighbourhood data, information about the locations, and additionally travel distance. The obtained results show that the supervised models generally perform well on predicting unique use, and visiting different cultural and sport locations. The models rely on a mix of aforementioned feature types, each varying in effect depending on the outcome. Based on these results, it can be concluded that machine learning can be an interesting tool in uncovering the underlying contribution of various factors in behaviour.
        URI
        https://studenttheses.uu.nl/handle/20.500.12932/33514
        Collections
        • Theses
        Utrecht university logo