decompose_m_crois

m_crois

<p>Entier indiquant la taille a partir de laquelle
on decide de "casser" la matrice ou non. Vaut 10 par defaut.</p>

taille_max

<p>Permet de decomposer totalement une matrice de croisement, qu'elle
soit connexe ou non, en plusieurs sous-matrices de croisements
connexes "incassables" ou ayant moins de lignes que <code>taille_max</code></p>

An algorithm based on graph theory tools to detect differentiation
problems. A differentiation problem occurs when aggregated
data are disseminated according to two different nomenclatures. By making the
difference for an additive variable X between an aggregate composed of categories
of the first nomenclature and an other aggregate, included in that first
aggregate, composed of categories of the second nomenclature,
it is sometimes possible to derive X on a small aggregate of records
which could then lead to a break of confidentiality.
The purpose of this package is to detect the set of aggregates composed of
categories of the first nomenclature which lead to a differentiation problem,
when given a confidentiality threshold. Reference: Vianney Costemalle (2019) <doi: 10.3233/SJI-190564>.

Vianney Costemalle

decompose_m_crois: Decomposer une matrice de croisement en composantes connexes

Description

Usage

Arguments