基于贝叶斯网络的大学生共享单车出行行为研究  

Research on college students’bike-sharing travel behavior based on Bayesian network

作　　者：刘诗序[1] 唐颖诺 王智煜 贺朝阳 LIU Shixu;TANG Yingnuo;WANG Zhiyu;HE Zhaoyang(College of Civil Engineering,Fuzhou University,Fuzhou,Fujian 350108,China;Patent Examination Cooperation Fujian Branch of Beijing Center of the Patent Office,National Intellectual Property Administration,Fuzhou,Fujian 350100,China)

机构地区：[1]福州大学土木工程学院,福建福州350108 [2]国家知识产权局专利局专利审查协作北京中心福建分中心,福建福州350100

出　　处：《福州大学学报（自然科学版）》2021年第1期26-32,共7页Journal of Fuzhou University(Natural Science Edition)

基　　金：国家留学基金资助项目(201806655005);国家自然科学基金资助项目(71804026)。

摘　　要：为研究大学生共享单车出行行为,以福州市大学城各高校学生为研究对象,利用问卷调查采集各年级学生共享单车出行数据.首先,基于所获得的数据计算各节点的互信息值,假设贝叶斯网络参数服从Dirichlet分布,采用K2算法进行贝叶斯网络结构学习,利用贝叶斯估计法进行贝叶斯网络的参数学习,从而构建大学生共享单车出行行为的贝叶斯网络.然后,利用所构建的网络进行共享单车出行方式预测,计算该模型的预测值与实际值的误差,分析模型的精度,且与常用的Logit模型预测结果进行比较.最后,在所构建的网络模型基础上,应用联合树引擎分析是否拥有私人交通工具、出行距离等影响因素对大学生共享单车出行行为的影响.分析结果表明,贝叶斯网络学习精度较高,比Logit模型预测结果更有效.In order to study the bike-sharing behavior of college students,data on bike-sharing travel of students of different grades in the survey area of Fuzhou University Town were collected.Assuming that the parameters of Bayesian network obey the Dirichlet distribution,the mutual information value of each node was calculated.The K2 algorithm was used to learn the Bayesian network structure and the Bayesian estimation was used to learn Bayesian network parameter,thus the Bayesian network of bike-sharing behavior of college students can be constructed.Then,based on the constructed networks,the bike-sharing travel mode was predicted,and the accuracy of the model was evaluated by the error between the predicted value and the actual value.The model was also compared with the Logit model.Finally,based on the constructed network model,a joint tree engine was used to analyze the impact of whether travelers have private transportation,travel distance,and other factors on the college students’bike-sharing travel behavior.The analysis results show that the learning accuracy of Bayesian network is higher than that of Logit model.It is more effective than Logit model.

关 键 词：共享单车 大学生 贝叶斯网络 出行行为 联合树 LOGIT模型 

分 类 号：TP399[自动化与计算机技术—计算机应用技术] U491[自动化与计算机技术—计算机科学与技术]