Dynamic Factorial Analysis by Means of Tucker3 Method
Jeannette Amaya & Pedro Nel Pacheco
Abstract
The Tucker3 method, proposed for Leyard Tucker
en 1966, identified between the techniques designed to data three-mode
analysis, is considered a generalization of the Principal Component Analysis
(PCA) and the Singular Value Decomposition (SVD), and use matricial management
and decomposition procedures to estimation of the model parameters and
correspond graphic representation in spaces of lower dimension to the original
data array. Thorough this method is possible to explain the information of a
three-mode data conjoint summarizing the entities associates for means of few
components, and to describe the possible interactions between the three sources
of variation in the data using one central array. In this paper to describe the
theoretic principles and the analytic supports of the Tucker3 method, which are
exemplified using hypothetical data.
Key words: Dynamic factorial analysis, Cubic data, Tucker3
model, Singular value decomposition, Factorial planes, Trajectories.
PDF (Spanish)