Bike-sharing systems are widely deployed in many major cities, providing a convenient transportation mode for citizens’ commutes. As the rents/returns of bikes at different stations in different periods are unbalanced, the bikes in a system need to be rebalanced frequently. Real-time monitoring cannot tackle this problem well as it takes too much time to reallocate the bikes after an imbalance has occurred. We propose a hierarchical prediction model to predict the number of bikes that will be rent from/returned to each station cluster in a future time interval so that bike reallocation can be executed in advance. We first propose a bipartite clustering algorithm to cluster bike stations into groups, formulating a 2-level hierarchy of stations. Then a H̲i̲erarchical T̲ime S̲eries...[ Read more ]

Bike-sharing systems are widely deployed in many major cities, providing a convenient transportation mode for citizens’ commutes. As the rents/returns of bikes at different stations in different periods are unbalanced, the bikes in a system need to be rebalanced frequently. Real-time monitoring cannot tackle this problem well as it takes too much time to reallocate the bikes after an imbalance has occurred. We propose a hierarchical prediction model to predict the number of bikes that will be rent from/returned to each station cluster in a future time interval so that bike reallocation can be executed in advance. We first propose a bipartite clustering algorithm to cluster bike stations into groups, formulating a 2-level hierarchy of stations. Then a H̲i̲erarchical T̲ime S̲eries (HiTS) prediction model based on an Input-output Hidden Markov Model (IO-HMM) and Gaussian Processes (GPs) is proposed to predict the check-out, from which the check-in of each cluster can be easily inferred. We evaluate our model on a real bike-sharing system in New York City (NYC), named Citi Bike, confirming our model’s advantage beyond baseline approaches, especially for anomalous periods, i.e. those under rare weather conditions.