Cluster-Based Destination Prediction in Bike Sharing System

Abstract

Destination prediction not only helps to understand users' behavior, but also provides basic information for destination-related customized service. This paper studies the destination prediction in the public bike sharing system, which is now blooming in many cities as an environment friendly short-distance transportation solution. Due to the large number of bike stations (e.g. more than 800 stations of Citi Bike in New York City), the accuracy and effectiveness of destination prediction becomes a problem, where clustering algorithm is often used to reduce the number of destinations. However, grouping bike stations according to their location is not effective enough. The contribution of the paper lies in two aspects: 1) Proposes a Compound Stations Clustering method that considers not only the geographic location but also the usage pattern; 2) Provide a framework that uses feature models and corresponding labels for machine learning algorithms to predict destination for on-going trips. Experiments are conducted on real-world data sets of Citi Bike in New York City through the year of 2017 and results show that our method outperforms baselines in accuracy.