基于LC-HB模型的铁路货运客户选择偏好差异性与市场细分方法

摘要

铁路货运客户自身的选择偏好直接影响对运输服务的选择，进而影响到不同货运服务的市场占有率，而选择偏好性受显性变量和隐性变量的影响。基于随机效用模型，分析铁路货运客户的运输服务选择模式，引入潜类模型，将客户选择偏好纳入先验概率和后验概率模型，并根据客户对服务属性的选择偏好差异性，建立铁路货运市场细分模型。同时，使用叙述性偏好实验，构建铁路货运服务效用模型，通过模拟服务选择过程，获取客户对不同货运服务的选择偏好数据。通过分层贝叶斯与潜类模型，对用户选择效用值和市场细分值进行迭代运算，得到实验样本选择偏好差异性的多元分布函数，及每个市场细分的客户规模、隶属度函数、对服务属性和服务层次的效用值。数据分析表明，客户选择偏好存在明显差异性，同时，解释并验证内生性市场细分方法能够对不同客户群体的隐性变量进行刻画，进而从铁路货运客户选择行为角度划分货运市场。%Railway shippers'choice behavior will be affected by their preferences on freight services,which are closely related to shippers'observable and unobservable variables,and choice behavior will further influence the market shares of different modes and services.Through Random Utility Model,the choice modes of railway shippers for services are analyzed.With the application of the Latent Class Model,shippers'choice preferences are included in both prior and posterior probability models.Based on the choice preference differentiation of shippers in connection with service attributes and levels,market segmentation models are established.Mean-while,a stated preference experiment was conducted to develop railway freight service utility model.Through the simulation of service selection process,shippers'choice preference data on various freight services are de-rived.Through Hierarchical Bayes and Latent Class estimation,shippers'choice utilities of different service at-tributes and the best market segmentation size are calculated by iteration process,to obtain multivariate distri-bution function,shipper size of each market segment,membership function,utilities of service attributes and levels in relation to the preference difference of the experiment sample.Results prove that significant difference exists in shipper preferences.At the same time,the endogenous segmentation method can describe the unob-servable variables of different shipper groups,thereby segmenting the freight market according to shippers' choice behavior.