Régression et analyse géométrique des données : réflexions et suggestions

摘要

Multivariate data are often treated with regression methods on one hand, Geometric Data Analyse methods (PCA, AC…) on the other hand. We intend to show, thanks to the mathematical structures common to the two methods, illustrated by examples, how one can integrate regression methods in geometric analysis. Geometric analysis allows a visualization of structural effects. There is no ground to oppose “explanatory” and “descriptive” statistical methods.