Central Station Based Demand Prediction in a Bike Sharing System

Predicting the bike demand can help rebalance the bikes and improve the service quality of a bike sharing system. A lot of work focuses on predicting the bike demand for all the stations. It is not necessary because the travel cost of rebalance operations increases sharply as the number of stations increases. In this paper, we take more attention to those stations with higher bike demand, which are called "central stations" in the following narrative. We propose a framework to predict the hourly bike demand based on the central stations we define. Firstly, we propose a novel clustering algorithm to assign different types of stations into each cluster. Secondly, we propose a hierarchical prediction model to predict the hourly bike demand for every cluster and each central station progressively. The experimental results on the NYC Citi Bike system show the advantages of our approach to these problems.