Factorisation de réseaux temporels : étude des rythmes hebdomadaires du système Vélo'v

We study the Vélo'v system, a fully automated bike-sharing system in Lyon, using a temporal network representation. A decomposition of this network is proposed by using nonnegative matrix factorisation (NMF), whose the choice of parameters is discussed. This decomposition enables us to represent the temporal network as a mixture of subnetwork, whose structures describe specific behaviors of Vélo'v users, in relation with the geographical and socio-economical space. Coefficients of activation associated to the patterns separate temporally each subnetwork, enabling an interpretation consistent with the literature on bike-sharing systems.