基于贝叶斯网络的城市居民出行方式研究

摘要

The trip modes of residents in Xi'an city were taken as study subject,and the survey data of some regions in 2013 were collected.A Bayesian network for trip mode analysis was developed by structure and parameter learning,using correlation analysis,K2 algorithm and Bayesian method.Based on the Bayesian network,the influences of private car,gen-der,age and travel purpose on the choice of trip mode were analyzed.The results show that the Bayesian network can ex-press the complicated relationship between the trip mode and the causes.Moreover,the Bayesian network has a high accura-cy.The study can contribute to the development of the trip behaviors theory of residents in Xi'an city.%以西安市城市居民出行方式为研究对象，收集西安市部分区域城市居民出行的调查数据。利用获得的调查数据，综合运用相关性分析方法和 K2算法进行贝叶斯网络的结构学习；应用贝叶斯参数估计方法进行贝叶斯网络的参数学习，建立了应用于西安城市居民出行方式分析的贝叶斯网络。应用所建网络分析了是否有私家车、居民性别、居民年龄和出行目的对西安城市居民出行方式的影响。研究结果表明，基于贝叶斯网络建立的西安城市居民出行方式分析模型预测精度较高，具有较高的实用价值。