RideMovi data: Analysis of bike flows

Cycling offers numerous benefits for urban environments and individuals alike. It is a sustainable and accessible mode of transport promoting carbon footprint reduction and public health enhancement. Moreover, well-developed cycling infrastructure is essential for the success of the 15-minute city. During recent years, multiple bike-sharing services keep facilitating accessibility to this mode of active mobility in urban areas. In order to improve bike-sharing services' user experience, it is important to know how bike flows change over time. For example, this information is vital for solving the Bike Spreading Problem. Normally, bike flow changes are periodic and exact mobility behaviors happen in specific periods. This periodic nature can be explained by different factors including the prevailing work schedule of people inhabiting the urban area. However, the way bike flows change, as well as the majority of the reasons behind these changes, are specific to the considered locality. Hence, a data-driven approach could be robust and suitable to highlight and analyze changing patterns of bike flows. In this thesis, bike flows analysis was carried out using the data provided by RideMovi bike-sharing service. The dataset contains information about anonymized bike rides over the last 1.5 years in the city of Padua. In order to retrieve frequent bikeflow patterns, a data-driven approach based on applying the K-means clustering algorithm to spatial matrices of initial and final positions was introduced in this work. Later, each obtained cluster was carefully analyzed, as well as the periodicity of its occurrence and clusters' distribution over time. Results on this approach have shown that observed clusters' properties and dynamics are consistent with mobility patterns that are very common in the city of Padua.