c7e1249ffc03eb9ded908c236bd1996d-Reviews.html

This paper addresses the problem of identifying the type of each agents from his/her partial preference data, in order to use this information to better estimate the underlying preferences for each type. The authors propose a Generalized RUM to model the behavior of such clustered agents. A reversible jump MCMC technique is used to estimate the latent variables, including the types of the agents. A theoretical analysis of the identifiability of the model and uni-modality of the likelihood posterior are presented. Quality There are three contributions of this paper.