Products characterization

Products'characterization starts with a simple description of the data. This allows to see global tendencies and sometimes aberrant data.
To do so, use the boxprod function which draws boxplots for each product and each descriptor.

Then products'characterization can be seen from a unidimensional or a multidimensional point of view.
A first function, decat, performs analyses of variance for each descriptor to see which ones are the most characteristic for all the products on the one hand and for which products in particular on the other hand. For a graphical visualization, use the barrow function to draw bar plots for each product.
From a multidimensional point of view, PCA or MFA can be performed and confidence ellipses drawn thanks to panellipse and panellipse.session function. These functions allows to see whether the products are well discriminated from each other and, when using panellipse.session, whether the panel is repeatable.

Description of the data

Unidimensional point of view

Multidimensional point of view