Structural learning of tree model for ordinal data
In this paper,the structural EM algorithm is used to learn the structure of the tree model of the ordinal data.More precisely,we assume that the ordinal variables originate from marginally discretizing a set of Gaussian variables.By learning the tree model of the latent Gaussian data,the tree model of the ordinal data is learned.The simulation results show that the structural EM algorithm is efficient in learning the ordinal tree structure.At the same time,this method is applied to a psychological example to analyze the relationship between the main characteristics that affect Myers-Briggs personality type.
tree modelstructure learningordinal datastructural EM algorithm
NETL
NSTL
万方数据
维普
