论文部分内容阅读
为刻画居住选址备选方案之间的空间相关性,选取通勤成本、住房成本、小区特征、家庭社会经济属性等效用变量,构建了基于广义极值理论的配对巢式Logit模型。利用2005年北京市居民出行调查数据,对模型参数进行估计和检验,并对模型进行弹性分析,研究了效用变量改变所引起的备选方案选择概率的改变。结果表明:配对巢式Logit模型具有比传统多项Logit模型更优的统计学特征,并且能刻画空间相关性随备选方案之间距离增加而衰减的特性;较之以往研究中仅假定相邻空间存在相关性,该模型更接近现实。
In order to characterize the spatial correlation between alternatives of residential locations, the utility model of commuting cost, housing cost, residential characteristics and family socio-economic attributes are selected to construct a matching nested Logit model based on generalized extreme value theory. Based on the 2005 Beijing resident travel survey data, the model parameters are estimated and tested, and the model is analyzed flexibly to study the change of the alternative probabilities caused by the change of utility variables. The results show that the matching nested Logit model has better statistical characteristics than the traditional multinomial Logit model and can characterize the spatial correlation decaying with the increase of the distance between alternatives. Compared with the previous study, There is a correlation between the space, the model is more realistic.