sdcMicro (version 2.6.3)

localSupp2: Local Suppression 2

Description

An Algorithm to perfom local suppression to achieve k-anonymity.

Usage

localSupp2(x, keyVars, w, importance=rep(1, length(keyVars)), method="minimizeSupp", k=1)

Arguments

x

data frame or matrix

keyVars

column index of key variables

w

column index of sampling weights

importance

weights for each key variable

method

minimizeSupp (default), further methods will be included in future versions of the package

k

parameter for k-anonymity.

Value

• Object from class localSupp2.
• xAnon
  resulting data with suppressions
• supps
  number of suppressions in the key variables
• totalSupps
  total number of suppressions.
• anonymity
  TRUE, if k-anonymity is achieved
• keyVars
  index of the key variables.
• importance
  weight vector for key variables
• k
  k for k-anoymity

Details

With the help of this algorithm you can achieve k-anonymity in an optimized way. The procedure set missings only to those key variables for which the importance is greater than 0. Key variables with higher importance will be prefered to be the variable which will used for suppression of specific values, i.e. the vector of importance assign to each key variables a weight which is considered by the algorithm. To guarantee k-anonymity the wrapper of function localSupp2 should be applied (localSupp2Wrapper()) However, if the importance of some key variables are equal to zero, the alorithm may not find a k-anonymity solution (because there isn't any solution reachable at all, for example). The easiest way to overcome this situation is to re-run the alorithm and allow for NA's in some more key variables, i.e. re-run the algorithm with importance greater than 0 for all entries of importance. This will result in k-anonymizated results and leads to only few suppressions in the key variables where the importance of the variables are considered. Method fastSupp avoids some calculation steps but this method is only significant faster if there is a large data sets with few key variables. However, fastSupp leads to an oversuppression (sligthly).

References

Templ, M. Statistical Disclosure Control for Microdata Using the R-Package sdcMicro, Transactions on Data Privacy, vol. 1, number 2, pp. 67-85, 2008. http://www.tdp.cat/issues/abs.a004a08.php

See Also

freqCalc, localSupp

Examples

## example from Capobianchi, Polettini and Lucarelli:
data(francdat)
l1 <- localSupp2(francdat, keyVars=c(2,4,5,6), w=8)
l1
l1$x
l2 <- localSupp2(francdat, keyVars=c(2,4,5,6), w=8, k=2)
l3 <- localSupp2(francdat, keyVars=c(2,4,5,6), w=8, k=4)

## long computation time, wait some seconds to get an information
## about the estimated computing time.
## l = localSupp2(free1, keyVars=1:3, w=30, k=2, importance=c(0.1,1,0.8))