基于网络模型的城市公共自行车需求量预测研究

自行车共享系统逐渐出现在许多城市中,由于在不同时间和站点的自行车需求量(租/还量)的不平衡,系统中各站点的自行车需要人工频繁地使其不断达到平衡状态,然而实时监控并不能很好的解决这个问题。因此,提出了一个基于网络图的预测模型,可以预测未来时间段内的某个站点自行车的需求量,提前对自行车的重新分配。首先,我们通过分层聚类算法对预测站点进行聚类,得到与其相关的站点簇。其次,对站点簇构建网络模型。最后,使用纽约和华盛顿两个自行车共享系统的数据进行实验和结果分析。结果发现同一簇的站点具有相似的使用模式,模型预测误差率不高于0.45,且能够应用于不同城市的自行车共享系统。

Research on demand prediction of Urban Bicycle Sharing based on network model

Bicycle-sharing systems are widely deployed in many major cities. As the rents/returns of bicycles at different stations in different periods are unbalanced, the bicycles in a system need to be rebalanced frequently. Real-time monitoring cannot tackle this problem well. Therefore, we propose a prediction model based on network diagram to predict the number of bicycles that will be rent from/returned to each station in a future period so that reallocation can be executed in advance. We first propose hierarchical clustering algorithm to cluster bike stations into groups-relevant station clusters. Then,network model of a relevant station cluster was constructed. Finally, we evaluate our model on two bicycle sharing systems in New York City (NYC) and Washington D.C. (D.C.) respectively. The results manifest that there is a similarity in the same cluster, model prediction error rate is not higher than 0.45, and can be applied to different urban bicycle sharing system.