A Dynamic Shared Bikes Rebalancing Method Based on Demand Prediction

Abstract

The free-floating bike sharing systems (BSSs) are booming all over the world. How to rebalance the bikes is a problem faced by all operators. To tackle this problem, firstly, we compare five models to predict shared bikes demand and choose time series and decision tree model. Then based on the prediction results, we propose a zone-based two-stage rebalancing model and an algorithm to solve this model. The proposed model divides the research area into two kinds of zones: zones with deficient bikes (ZDB) and zones with sufficient bikes (ZSB). The objective of the model is to optimize the matching degree of the demand and actual number of shared bikes in each zone. Finally, we employ real world data to validate the flexibility and practicality of our model and algorithm. Experimental results demonstrate that this method can effectively balance all zones.