论文部分内容阅读
为了准确把握尺度参数在模型建模、分析以及应用过程中的作用,以基于极值分布的Multinomial Logit模型和Nested Logit模型为研究对象,从建模的基础——误差项的分布形式入手,通过模型形式和误差项性质两方面研究了尺度参数在两种模型中的特性,并对其进行了对比分析。分析结果表明:尺度参数不仅限制了模型效用函数的尺度,而且也反映了误差项的方差水平,但对于以上两种不同的模型,其在尺度限制、误差项方差水平及独立性、自身取值范围等方面均表现出不同的特性,Multinomial Logit模型的尺度参数反映了有关整个效用确定项的误差项方差水平,而Nested Logit模型的尺度参数只是反映了与某一选择枝有关的部分效用确定项的误差项方差水平。
In order to accurately grasp the role of scale parameters in model modeling, analysis and application process, taking Multinomial Logit model and Nested Logit model based on extreme value distribution as the research object, starting from the distribution of error model The form of the model and the nature of the error term, the characteristics of the scale parameter in the two models are studied and compared. The analysis results show that the scale parameter not only limits the scale of the model utility function, but also reflects the variance level of the error term. However, for the above two different models, the scale limit, the variance level and independence of the error term, Scale and other aspects of the Multinomial Logit model reflect the error variance variance of the whole utility determination item, while the scale parameter of the Nested Logit model only reflects the partial utility determination items related to a selection branch The variance of the error item level.