Les matrices inverses généralisées et leur utilisation dans le modèle linéaire

In this note, we first present three different types of generalized inverse matrices (conditional, least squares and MOORE-PENROSE inverse matrices), as well as some of their properties. Applications of these inverse matrices in the linear model are then examined (normal equations solutions, orthogonal projections, sums of squares and their sampling distributions, estimations of parametric functions and sums of squares related to hypotheses testing).