Search

Note, Bikesharing systems (BSS) are becoming increasingly popular in urban areas around the world, as demonstrated by the rapid growth of both the number and the size of these systems in recent years. Understanding and predicting BSS usage patterns is complex, especially because these patterns are often tied to local factors. This thesis aims to contribute to the existing literature on BSS in two ways. First, an econometric model featuring bicycle availability at a station level as a direct metric of analysis is developed. This behaviorally quantitative model accounts for the influence of temporal, meteorological, bicycle infastructure, built environment and land-use attributes on bicycle availability. More specifically, an ordered regression model-panel mixed generalized ordered logit model-is estimated to accommodate for the influence of exogenous variables and station level unobserved factors. The model estimation is undertaen using BIXI-Montreal data from the summer of 2012. The results show BIXI is used more in the afternoon than in the morning, dense areas tend to be associated with lower availability levels, and interactions of time of day with land use impact avialability. The estimated model is validated using a old-out sample of data from the summer of 2013. The results clearly highlight the satisfactory performance of the proposed framework. The model developed can be employed by BSS operators to arrive at hourly system state prediciton and used for rebalancing operations. To illustrate its applicability, an availability prediction exercise is also undertaken. Second, a BSS model transferability exercise is conducted using data from Montreal - to data from New York and comparing the influence of temporal, meterological, bicylce infastructure, built environment and land-use variables on BSS usage in these two cities. Results show significant overlap in the influence of weather variables, bicycle infastructure, and several land-use attributes. However, temporal t, Les Systemes de Velo en Libre-Service (SVLS) sont de plus en plus populaires dans les regions urbaines partout dans le monde, comme le demontre l'expansion de ces systemes au cours de la derniere decennie, en termes de nombre de SVLS et de leur taille. Les motifs d'utilisation des SVLS sont complexes et difficiles a predire, particulierement parce que ces motifs sont souvent lies a des facteurs locaux. Cette these vise a contribuer a la litterature academique sur les SVLS de deux manieres. Premierement, modele econometrique presentant la disponibilite de velos dans les stations BIXI comme metriue d'analyse directe est developpe. Ce modele comportemental quantitatif prend en compte l'influence de donnees temporelles, meteorologiques, de l'infastructure pour cyclistes, de l'infastructure denerale et de la gestion du territoire sur la disponibilite des velos. Plus particulierement, un modele de regression ordonne-modele logit panel mixte generalise ordonne-est estime afin d'accomoder l'influence de donnees exogenes ainsi que les facteurs non-observes au niveau des stations. Le modele est valide avec des donnees de l'ete 2013. Les resultats demontrent clairement la performance satisfaisante de la structure statistique poroposee. Le modele developpe peut etre employe par les operateurs de SVLS afin d'arriver a des predictions de disponibilite a une haute resolution temporelle-heure par heure-et peut etre utilise pour optimiser les operations de reequilibrage. Afin d'illustrer ces applications, un exercice de prediction de disponibilite est presente.