公共交通方式对居民出行影响的Mix-Logit测度研究

Impact of Public Transport Properties on Residents Utility based on the Mix-Logit Model

目的:为了探明居民公共出行特征,特别是公共交通方式特性对居民出行行为的影响程度。方法:以武汉市为例开展了一次居民公共出行意愿调查。结合武汉市公共交通现状,采用离散选择法设计调查问卷,开展基础数据调研,回收了490份有效问卷。首先,介绍了调查指标的选取、数据的处理与校对、出行效用Mix-Logit模型的待估参数;然后,对有效样本进行吉布斯采样生成马尔科夫链,用Metropolis-Hastings算法做蒙特卡洛仿真,从而获得一个符合现实出行选择的多元分布,采用双层贝叶斯估计法完成模型参数估计;最后,采用逐步回归法进行模型的检验。结果:经过3次逐步回归分析,得到了最大的回归方程显著性检验指数27.886,判断系数为0.738、拟合优度为0.766,说明模型参数标定较好。结论:居民出行费用的测度是出行时间的2.6倍,是出行舒适度的1.6倍,说明出行者敏感度最低的是出行费用属性。该模型可以应用在交通诱导方案制定、交通信息发布等领域,也可为公交(地铁)限时分阶段计价提供参考。

Objective:A survey of Wuhan residents’public willingness to travel is conducted to know about their public travel relevant characteristics,especially how the dynamic variation of public transport properties influences residents’travel behavior.Method:Based on the public transportation in Wuhan,discrete choice approach is applied to design questionnaire and carry out the basic data research,finally 490 valid questionnaires can be recycled.After introducing the choice of survey indicators,data processing and proofreading and parameters to be estimated of traveling efficacy Mix-Logit model,Gibbs sampling on valid samples to generate Markov chain,to obtain a multivariate distribution of realistic travel choice by using Metropolis-Hastings algorithm to progress Monte Carlo simulation,double Bayesian estimation method is used to complete model parameters estimation,finally using the stepwise regression method for inspection of the model.Results:The results show that the judgment coefficient is 0.738 and the goodness of fit is 0.738,which explains that model parameter calibration is better.The greatest significance test index regression equation 27.886 is obtained after 3 stepwise regression analyses.Conclusion:The coefficient of the property(cost of travel)is the largest,which shows that travel cost is the minimum sensitivity of traveler,measure of residents travel cost is 2.6 times of travel time and 1.6 times of travel comfort based on parameter calibration.The model not only can be applied in the field of traffic induction plan formulation and traffic information release etc.,but also can be the reference for bus(or subway)limit pricing.