摘要： 共享单车的需求量预测是优化车辆系统布局、实现车辆合理调度的基础。为了提高共享单车需求量预测模型的精度，建立了基于格兰杰因果分析和相似日选择的组合预测模型，研究了时间和天气因素对共享单车出行需求的影响。应用格兰杰因果检验方法，筛选出影响共享单车需求量变化的关键天气指标。然后，基于天气特征向量的灰色关联度指标，提取待预测日各时段的相似日样本集。综合随机森林回归、支持向量回归等机器学习算法，建立了 Stacking 策略的组合预测模型，对区域分时共享单车需求量进行预测。最后，对北京市共享单车用户的骑行数据进行实例分析。结果表明相较单个机器学习预测模型，提出的组合预测模型的平均绝对百分比误差下降了 9.1% ，提高了共享单车短时需求预测的科学性和准确性，可为实际车辆调度提供参考依据。

Abstract: Short-term demand of shared bikes forecasting plays an important role in optimizing the layout of free-floating bike sharing systems and bikes rebalancing. To improve the accuracy of demand forecasting methods for the emerging shared bikes business, this study establishes a combined forecasting model based on Stacking strategy and examines the impact of temporal variables and weather factors on shared bikes demand. In particular, the Granger causality test is used to identify the key weather indicators that cause demand fluctuations. We extract the set of similar samples for each period of the day for prediction based on the grey correlation index of weather variables. The Stacking strategy is then introduced to integrate random forest, support vector regression, and other machine learning algorithms for establishing a combined forecasting model to predict the short-term demand in different regions. Finally, using free-floating bike sharing data in Beijing, the proposed combined model is tested. The prediction results of combined model demonstrate that the mean absolute percentage error decreased by 9.1% compared with the single prediction model, thus improving the accuracy of the short-term demand forecast of shared bikes and providing useful information for bike relocation.

Key words: shared bikes, travel demand , causal analysis , grey correlation , similar day , machine learning , Stacking strategy BAO J, YU H, WU J M. Short-term FFBS demand prediction with multi-source data in a hybrid deep learningframework[J]. The Institution of Engineering and Technology,2019,13(9)：1340-1347. doi: 10.1049/iet-its.2019.0008 孔静. 无桩式共享单车站点需求预测及调度路径优化研究[D].西安：长安大学,2018. ZHOU Y J, WANG L L, ZHONG R,et al. A markov chain based demand prediction model for stations in bike sharing systems[J].Mathematical Problems in Engineering, 2018, 2018:1-8. doi: 10.1155/2018/8028714 KALTENBRUNNER A, MEZA R, GRIVOLLA J, et al. Urban cycles and mobilitypatterns:exploring and predicting trends in a bicycle-based public transport system[J].Pervasive & Mobile Computing, 2010, 6(4):455-466. doi: 10.1016/j.pmcj.2010.07.002 闫厦. 基于站点聚类的公共自行车系统需求量预测[D]. 大连：大连理工大学,2018. MATTSON J, GODAVARTHY R. Bike share in Fargo, North Dakota: keys to success and factors affectingridership[J]. Sustainable Cities and Society, 2017, 34:174-182. doi: 10.1016/j.scs.2017.07.001 CAMPBELL A A,CHERRY C R,RYERSON M S, et al.Factors influencing the choice of shared bicycles and shared electric bikes in Beijing[J]. Transportation Research Part C: Emerging Technologies,2016,67. doi: 10.1016/j.trc.2016.03.004 RIXEY R A. Station-level forecasting of bike sharing ridership: station network effects in three U.S. Systems[M]. London:SAGE Publications,2012. doi: 10.3141/2387-06 CHEN L B, J J, ZHANG D Q , et al. Dynamic cluster-based over-demand prediction in bike sharing systems[C]//Acm International Joint Conference on Pervasive & UbiquitousComputing.NewYork:ACM,2016:841-852. doi: 10.1145/2971648. 2971652 种颖珊,韩晓明.基于随机森林与时空聚类的共享单车站点需求量预测[J].科学技术与工程,2018,18(32):89-94. doi: 10.3969/j.issn.1671-1815.2018.32.015 LI Y X,ZHENG Y,ZHANG H C, et al. Traffic prediction in a bike-sharing system[C]// Proceedings of the 23rd ACM International Conference on Advances in Geographical Information Systems. Bellevue:ACM Sigspatial, 2015:1-10. doi: 10.1145/2820783.2820837 CHANG X M, WU J J, HE Z B, et al. Understanding user's travel behavior and cityregion functions from station-free shared bike usage data[J]. Transportation Research Part F: Traffic Psychology and Behaviour,2020,72:81-95. 周志华.机器学习及其应用[M]. 北京：清华大学出版社,2007. CANI P D, DEPOMMIER C. Reply to Simpson's paradox in proof-of-concept studies'[J]. Nature medicine,2019,25:1640-1641. doi: 10.1038/s41591-019-0625-x 任伟杰，韩敏.多元时间序列因果关系分析研究综述[J/OL].自动化学报:1-15.（2020-04-13）[2020-07-23].https://doi.org/10.16383/j.aas.c180189. 吴潇雨,和敬涵,张沛,等.基于灰色投影改进随机森林算法的电力系统短期负荷预测[J].电力系统自动化,2015,39(12):50-55. doi: 10.7500/AEPS20140916005 IBOLD S, NEDOPIL D C,The Evolution of free-floating bike-sharing in China[EB/OL].(2018-08-03)[2020-07-23].https://www.sustainabletransport.org/archives/6278.