Abstract

CUARITA AJNO, Lucy Gabriela. Descomposición tensorial Tucker3 aplicado a tablas de contingencias de tres vías. Revista Varianza [online]. 2020, n.17, pp. 20-28. ISSN 9876-6789.

Abstract Tensor data analysis is responsible for the study of data obtained from the measurement of more than one variable on a set of individuals or objects, which are arranged in a higher order tensor and where the decomposition of the tensor into much more components is of fundamental interest. simple, in such a way that they facilitate the interpretation of the data. In the field of Multivariate Analysis, in particular, the Multiple Correspondence Analysis technique allows identifying the interaction of the levels corresponding to a set of study variables, transforming the contingency table and then applying the Simple Correspondence Analysis technique. On the other hand, the Tucker3 tensor model is a tensor decomposition method that allows modeling the interaction between the pathway s of a third order tensor and its components, preserving the original structure of the data. Today, tensor models are an alternative in multivariate data analysis, although most of the work is in the field of three-way data analysis, there is research that indicates that the methodology will continue to rise as long as the structures of data become increasingly complex and researchers require a comprehensive analysis of the data.

Keywords : Pearson 's statistic; Inertia; Interaction; Tensor Decomposition; Tucker3 Model; Tensor.