PTxyz: ProtectTable with output ready for SuppressDec in package...

View source: R/easySdcTableGauss.R

PTxyzR Documentation

ProtectTable with output ready for SuppressDec in package RegSDC

Description

Assuming correct suppression, suppressed values become decimal numbers (not whole numbers) instead of missing.

Usage

PTxyz(data, dimVar, freqVar, ...)

Arguments

data

data frame

dimVar

The main dimensional variables and additional aggregating variables (name or number).

freqVar

Variable(s) holding counts (name or number).

...

Further parameters sent to ProtectTable

Details

Within this r package this function will be used for testing

Value

List of three matrices ready as input to SuppressDec

x

Sparse dummy matrix where the dimensions match z and y.

z

Frequencies to be published with suppressed as NA.

y

Inner cell frequencies.

Author(s)

Øyvind Langsrud

Examples

## Not run: 

# Same examples as in ProtectTable 
a1 <- PTxyz(EasyData("z1"), c("region","hovedint") ,"ant")
a2 <- PTxyz(EasyData("z2"), c(1,3,4),5) 
 
if (require(RegSDC)) { # RegSDCdata and SuppressDec
  # Same data as in RegSDCdata examples (and in paper)
  data7 <- RegSDCdata("sec7data")
  data7 <- data7[!is.na(data7$y), 1:3]
  data7

  # Generate x, y, z similar to xAll, y, zAllSupp in RegSDCdata examples 
  # But different suppressed cells and z ordered differently
  a <- PTxyz(data7, 1:2, 3, maxN = 3, method = "HITAS")
  a

  # Suppressed inner cells as decimal numbers
  yDec <- SuppressDec(a$x, a$z, a$y, rmse = 1)
  yDec

  # All cells as decimal numbers
  cbind(a$z, t(a$x) %*% cbind(a$y, yDec))

  # As ProtectTable example
  z1 <- EasyData("z1")
  a <- PTxyz(z1, c("region", "hovedint"), "ant")

  # Inner cells as decimal numbers. 3 replicates.
  yDec <- SuppressDec(a$x, a$z, a$y, nRep = 3)
  yDec

  # All cells with 3 replicates.
  cbind(a$z, t(a$x) %*% cbind(a$y, yDec))
}

if (require(RegSDC)) {
  # An example involving two linked tables.  
  # It is demonstrated that the SIMPLEHEURISTIC approach to suppression is not safe.  
  # That is, perfect fit (whole numbers) for some suppressed cells.
  a <- PTxyz(EasyData("z3"), 1:5, 7, method = "SIMPLEHEURISTIC", protectZeros= FALSE)
  cbind(a$z, t(a$x) %*% cbind(a$y, SuppressDec(a$x, a$z, rmse=pi/3, nRep=3)))[which(is.na(a$z)), ]
}
## End(Not run)

statisticsnorway/Kostra documentation built on Nov. 2, 2024, 6:40 p.m.