J. F. Vera Vera
Distance association (DA) models have become a generally useful tool for the analysis of cross-classified data, allowing the interpretation of the relationship present between row and column categories in terms of Euclidean distances. Nevertheless, the presence of a large amount of modalities and/or zeros in tables with large number of cells makes difficult the interpretation. Co-clustering in combination with DA models results a useful tool for the analysis of such large and sparse datasets. A model based clustering procedure in combination with a generalized EM algorithm is proposed. The model makes cluster in the rows and in the columns of a contingency table while simultaneously represents both cluster centres in a low dimensional space using Unfolding. An estimation procedure using the GEM algorithm is employed for the analysis of the association between personality variables plus gender with personality disorders.
Palabras clave / Keywords: co-clustering, distance association, unfolding
Programado
Sesión GT02-4: Análisis Multivariante y Clasificación (AMyC-4). Organizadora: Eva Boj del Val
29 de mayo de 2018 17:00
Sala 5
