addNoise {sdcMicro} | R Documentation |
Various adding noise methods for the perturbation of data can be used.
addNoise(x, noise = 150, method = "additive", p = 0.001, delta=0.1)
x |
data frame or matrix which should be perturbed |
noise |
amount of noise |
method |
choose between ‘additive’, ‘correlated’, ‘correlated2’, ‘restr’, ‘ROMM’, ‘outdect’ |
p |
multiplication factor for method ‘ROMM’ |
delta |
parmeter for method ‘correlated2’, see in the reference below |
Method ‘additive’ adds noise completely random to each variable depending on there size and standard standard deviation. ‘correlated’ and method ‘correlated2’ adds noise and preserves the covariances as descriped in R. Brand (2001) or in the reference given below. ‘restr’ takes the sample size into account when adding noise. Method ‘ROMM’ should be the implementation of the first algorithm of Fienberg (2004). Method ‘outdect’ adds noise only to outliers. This outliers are idedentified with univariate and robust multivariate procedures based on a robust mahalanobis distancs calculated with the MCD estimator.
An object of class “micro” with following entities:
x |
the original data |
xm |
the modified (perturbed) data |
method |
method used for perturbation |
noise |
amount of noise |
Matthias Templ
Domingo-Ferrer, J. and Sebe, F. and Castella, J., “On the security of noise aliition for privacy in statistical databases”, Lecture Notes in Computer Science, vol. 3050, pp. 149-161, 2004. ISSN 0302-9743. Vol. Privacy in Statistical Databases, eds. J. Domingo-Ferrer and V. Torra, Berlin: Springer-Verlag. http://vneumann.etse.urv.es/publications/sci/lncs3050OntheSec.pdf,
Ting and Fienberg, “ROMM Methodology for Microdata Release” Joint UNECE/Eurostat work session on statistical data confidentiality, Geneva, Switzerland, 2005, http://www.niss.org/dgii/TR/wp.11.e(ROMM).pdf
data(Tarragona) a1 <- addNoise(Tarragona) a1 valTable(Tarragona[100:200,], method=c("addNoise: additive", "addNoise: correlated2"))