A Tucker3 method application on adjusted-PMRs for the study of work-related mortality

Principal Component Analysis is a widely used technique on two-way matrices for both dimensionality reduction and interpretation of latent relations among observed variables. The Tucker3 method is a generalization of PCA for three-way matrices, which not only runs classical PCA on each mode (way) of the data but also gives an estimate of the interrelation among the three modes. In the current analysis, the Tucker3 method is applied on data concerning mortality in the male population of Italy, in 2005-2015, specified for different causes of death, in people who have been working in different sectors and have had different levels of education; the main goal of this analysis is to understand the underlying relations among the three variables, in the considered population.