Intertemporal Community Detection in Bikeshare Networks
We investigate the changes in the patterns of usage in the Divvy bikeshare system in Chicago from 2016-2018. We devise a community detection method that finds clusters of nodes that are increasing, decreasing, or stable in connectivity across time. We use an iterative testing approach that is augmented by trend testing and a novel temporal false-discovery-rate correction. Results show stark geographical patterns in clusters that are growing and declining in relative bike-share usage across time and may elucidate latent economic or demographic trends.
READ FULL TEXT