Abstract
This paper presents the results of analyzing the time load of stations in bike-sharing systems using temporal networks. Temporal networks have many applications in the study of the behavior of complex dynamic systems that have a network structure. In particular, they can be used to analyze and predict many dynamic indicators of transport networks, for example, such as the intensity of transport and passenger flows, traffic congestion, capacity of transport nodes, turnover of vehicles, etc. In this work, the indicators of the centrality of stations and clusters of a bike-sharing network are estimated using temporal networks. Based on the obtained estimates, visual models (Heat maps and Time Series) are constructed to demonstrate the spatial and temporal features of the bike network in a clear and compact form. The station centralities are estimated on the basis of the betweenness measure, and the cluster centralities are estimated on the basis of the Freeman centralization. Experiments confirming the applicability of the built models are conducted using open data from the CitiBike New York system for April 2019. They demonstrated the presence of daily and monthly patterns among both individual stations and more large station clusters.