We study the effect of customer choices in bicycle-sharing systems based on bicycle availability predictions. We show that such systems may lead to flapping behavior between bicycle stations. The consequences of flapping instability include poor user experience and suboptimal usage of the available bicycle stock. We propose a simple assignment strategy aimed at eliminating flapping and balancing demand at each station based on actual availability.
On Closed-Loop Bicycle Availability Prediction
IEEE Transactions on Intelligent Transportation Systems ; 16 , 3 ; 1449-1455
2015-06-01
545557 byte
Article (Journal)
Electronic Resource
English