In this contribution we provide initial findings to the problem of modeling fuzzy rating responses in a traditional probabilistic context. In particular, we study a probabilistic tree model with the aim of representing the stage-wise mechanisms of direct fuzzy rating scales. A Multinomial model coupled with a mixture of Binomial distributions is adopted to model the parameters of LR-type fuzzy responses whereas a binary decision tree is used for the stage-wise rating mechanism. Parameter estimation is performed via marginal maximum likelihood approach whereas the characteristics of the proposed model are evaluated by means of an application to a real dataset.
翻译:在这一贡献中,我们对在传统概率范围内模拟模糊评级答复的问题提供了初步结论,特别是,我们研究了一种概率树模型,目的是代表直接模糊评级尺度的分阶段机制。我们采用了一种多数值模型,加上二相配的二相配,以模拟LR型模糊响应的参数,而二相决策树用于分阶段评级机制。参数估算是通过最低可能性方法进行的,而拟议模型的特征则通过对真实数据集的应用进行评估。