Cluster analysis of parking behaviour: A case study in Munich

Estimates show that vehicles cruising for on-street parking contribute to 30% of urban traffic congestion. On-street parking information (OSPI) systems are increasingly becoming a more popular service to help lessen the on-street parking search time and consequently reduce congestion. However, despite the service offerings of these prediction models, the on-street parking behaviour of people in cities have not been studied to the same magnitude. The lack of appropriate empirical parking data is one main reason. This study focuses on the analysis of parking behaviour by capturing the on-street parking dynamics, which can give a better insight on a city's parking contextualization. The case study examined is the parking behaviour dynamics within Munich by inferring from parked-in and parked-out events data from vehicles. A two part clustering analysis was conducted: (1) agglomerative clustering on the temporal trend of parking dynamics (TTPD) and (2) a two-stage DBSCAN-K-means clustering on the parking duration information. The results show that using the methodology introduced, the parking behaviour within the city can be obtained using this unsupervised learning approach.