| 4,803 | 116 | 330 |
| 下载次数 | 被引频次 | 阅读次数 |
基于共享单车项目的多维度大样本数据,以套索回归、岭回归、随机森林和迭代决策树等机器学习模型,探讨了共享单车短期(基于小时)需求预测的主要影响因素,并对不同模型预测效果进行了比较分析。研究结果发现,影响共享单车小时需求的主要因素包括特定的位置因素、时间因素以及天气条件因素。同时,相比普通线性回归、套索回归和岭回归模型,随机森林和迭代决策树模型对共享单车短期即时需求预测的结果更精确,在样本内部拟合和样本外推预测中的拟合优度(R2)更高,标准误差(RMSE)更低,是共享单车行业短期实时需求精准预测的更有效手段。
Abstract：Using the large and multidimensional data released by the bike-sharing project,and employing the Machine Learning Models,this article discussed the factors influenced short-term demand prediction of a bike-haring business. The results showed that the major factors that affected the short-term demand of bike sharing include specific location,time,and weather conditions. Meanwhile,compared with the Ordinary Linear Regression,Lasso Regression and Ridge Regression model,the Random Forest and Gradient Boosting Decision Tree models had higher goodness of fit( R2) and lower standard error( RMSE) in both in sample and the out sample predictions,which shed lights on the machine learning models and are more suitable for short-term precise demand predictions.
[1]MEDDIN R,DEMAIO P.The bike sharing world map[EB/OL].(2007-01-01)[2017-09-10].http://www.metrobike.net/the-bike-sharing-world-map/.
[2]DEMAIO P.Bike-sharing:history,impacts,models or provision and future[J].Journal of Public Transportation,2009,12(4):41-56.
[3]CAMPBELLA,CHERRY C,RYERSON M,et al.Factors influencing the choice of shared bicycles and shared electric bikes in Beijing[J].Transportation Research Part C,2016,67(6):399-414.
[4]MATTSON J,GODAVARTHY R.Bike share in Fargo,North Dakota:keys to success and factors affecting ridership[J].Sustainable Cities and Society,2017,34(10):174-182.
[5]FAGHIH-IMANI A,ELURU N,EL-GENEIDY A,et al.How landuse and urban form impact bicycle flows:evidence from the bicycle-sharing system(BIXI)in Montreal[J].Journal of Transport Geography,2014,41(12):306-314.
[6]NOSAL T,MIRANDA-MORENO.The effect of weather on the use of North American bicycle facilities:a multi-city analysis using automatic counts[J].Transportation Research Part A,2014,66(8):213-225.
[7]GEBHART K,NOLAND R.The impact of weather conditions on bikeshare tirps in Washington D C[J].Transportant,2014,41(6):1205-1225.
[8]FAGHIH-IMANI A,HAMPSHIRE R,MARLA L,et al.An empirical analysis of bike sharing usage and rebalancing:evidence from Barcelona and Seville[J].Transportation Research Part A,2017,97(3):177-191.
[9]RIXEYR.Station-level rorecasting of bikesharing ridership:station network effects in three U.S.systems[J].Transportation Research Record:Journal of the Transportation Research Board,2013,2387(1):46-55.
[10]WANG X,LINDSEY G,SCHONER J,et al.Modelling bike share station activity:effects of nearby businesses and jobs on trips to and from stations[J].Journal of Urban Planing and Development,2016,142(1):1-9.
[11]EL-ASSI W,MAHMOUD M,HABIB K.Effects of built environment and weather on bike sharing demand:a station level analysis of commercial bike sharing in Toronto[J].Transportation,2017,44(3):589-613.
[12]FISHMAN E,WASHINGTON S,HAWORTH N,et al.Barriers to bikesharing:an analysis from Melbourne and Brisbane[J].Journal of Transport Geography,2014,41(12):325-337.
[13]COCK J.Bike share in small and medium-sized cities[C].Washington D C:Presentation at 2016 Transportation Research Board Tools of the Trade Conference,2016.
[14]TANG Y,PAN H,FEI Y.Research on user's frequency of ride in Shanghai Minhang Bikesharing System[C].Shanghai:World Conference on Transport Research,2016.
[15]KASPI M,RAVIV T,TZUR M.Detection of unusable bicycles in bike-sharing systems[J].The International Journal of Management Science,2016,65(12):10-16.
[16]EINAV L,LEVIN J.Economics in the age of big data[J].Science,2014,346(11):715-721.
[17]MITCHELL T.Machine learning[M].New York:Mc Graw Hill,1997:2.
[18]MULLAINATHAN S,SPIESS J.Machine learning:an applied econometric approach[J].Journal of Economic Perspectives,2017,31(2):87-106.
[19]BAJARI P,NEKIPELOV D,RYANS P,et al.Machine learning methods for demand estimation[J].American Economic Review,2015,105(5):481-485.
[20]BELLONI A,CHERNOZHUKOV V,HANSEN C.High-dimensional methods and inference on structural and treatment effects[J].Journal of Economic Perspectives,2014,28(2):29-50.
[21]BACCIU D,CARTA A,GNESI S,et al.An experience in using machine learning for short-term predictions in smart transportation systems[J].Journal of Logical and Algeraic Methods in Programming,2017,87(2):52-66.
[22]刘涛雄,徐晓飞.大数据与宏观经济分析研究综述[J].国外理论动态,2015(1):57-64.
[23]曾照云,程晓康.德尔菲法应用研究中存在的问题分析——基于38种CSSCI(2014-2015)来源期刊[J].图书情报工作,2016(16):116-120.
[24]刘卫东,仲伟周,石清.2020年中国能源消费总量预测——基于定基能源消费弹性系数法[J].资源科学,2016(4):658-664.
[25]张钠,殷哲,吕飞,等.面向连锁零售业的时间序列预测方法研究[J].2014(3):60-65,71.
[26]刘涛雄,徐晓飞.互联网搜索行为能帮助我们预测宏观经济吗?[J].经济研究,2015(12):68-83.
[27]何国华.区域物流需求预测及灰色预测模型的应用[J].北京交通大学学报(社会科学版),2008(1):33-37.
[28]HOERL A E,KENNARD R W.Ridge regression:biased estimation for non-orthogonal problem[J].Technometrics,1970,12(1):55-88.
[29]TIBSHIRANI R J.Regression shrinkage and selection via the lasso[J].Journal of the Royal Statistical Society,1996,58(1):267-288.
[30]BREIMAN L,FRIEDMAN J,OLSHEN R,et al.Classification and regression trees[M].London:Chapman&Hall,1984:14-55.
[31]VARIAN H R.Big data:new tricks for econometrics[J].Journal of Economic Perspectives,2014,28(2):3-28.
[32]BREIMAN L.Random forests[J].Machine Learning,2001,45(1):5-32.
[33]CHEN T,GUESTRIN C.XGBoost:a scalable tree Boosting System[EB/OL].(2016-03-09)[2017-09-15].ar Xiv:submit/1502704.http://arxiv.org/abs/1603、02754v3.
[34]ATHEY S,IMBENS G.The state of applied econometrics:causality and policy evaluation[J].Journal of Economic Perspectives,2017,31(2):3-32.
[35]FELSON M,SPAETH J L.Community structure and collaborative consumption:a routine activity approach[J].American Behavioral Scientist,1978,21(4):614-624.
(1)共享经济概念由Felson和Spaeth(1978)首次提出[35],主要特点是个体通过第三方平台实现点对点(Peer to Peer)的直接交易。目前对共享经济的概念仍存不同理解。本文将自行车的分时租赁和共享服务统称为“共享单车”。
(2)按照DeM aio(2009)的总结,在现代互联网技术应用到单车租赁行业之前,共享单车在技术和模式上经历过三代发展,包括1965年首次在阿姆斯特丹出现的第一代“白色单车”(White Bikes),第二代1995年在丹麦迅速发展的“城市单车”(City Bikes)和第三代1996年在英国出现的磁卡单车。目前,以互联网、定位技术、移动设备、网络支付为基础的新一代共享单车可称为第四代。
(1)更详细的信息可见http://kalw.org/post/sf-bay-area-bike-share-launches-thursday#stream/0。
(2)读者可以在https://www.kaggle.com/benhamner/sf-bay-area-bike-share上获取相关数据。
(1)最终纳入模型的有法定假期的节日每年有17天,分别按照样本对应的日期加入虚拟变量。
基本信息:
DOI：10.14134/j.cnki.cn33-1336/f.2018.08.002
中图分类号:F572;F724.6
引用信息:
[1]焦志伦,金红,刘秉镰,等.大数据驱动下的共享单车短期需求预测——基于机器学习模型的比较分析[J].商业经济与管理,2018,No.322(08):16-25+35.DOI:10.14134/j.cnki.cn33-1336/f.2018.08.002.
基金信息:
国家自然科学基金项目“考虑消费者行为的O2O服务企业决策优化与供应链协同研究”(71772095); “中国特色社会主义经济建设协同创新中心”项目支持; 南开大学人文社会科学青年教师研究启动项目“互联网革命与物流业态变革研究”
2018-09-07
2018-09-07
2018-09-07