Home / Questions / 525846
Accepted
Patricio Moracho
Asked: 2022-09-04 14:42:22 +0800 CST 2022-09-04 14:42:22 +0800 CST

How to anonymize strings from a data.frame?

  • 772

I need to share a data.framewith a colleague, but would like to somehow "anonymize" the data. The idea would be:

  • Nothing too advanced (I don't need to adhere to any standards or norms)
  • not reversible
  • only for chains
  • simple and fast

Suppose some data like this:

df <- data.frame(nombre = c('Juan', 'Pedro'),
                 Edad = c(34, 45),
                 dni = c('12345678', '87654321'))

The idea would be to apply the algorithm only on the name and the ID.

r

1 Answers

  1. Best Answer

    I can think of a few alternatives:

    1. Transform the strings into a factor and then use the index of eachlevel

    anonymize_str_vector <- function(x) {
  if( !is.character(x) ) stop('x is not a vector of characters')
  as.character(as.numeric(factor(x)))
}

anonymize_str_df <- function(df) {
  if( !is.data.frame(df) ) stop('df is not a data.frame')
  convert <- sapply(df, is.character)
  df[,convert] <- sapply(df[,convert], anonymize_str_vector)
  df
}

anonymize_str_df(df)

  nombre Edad dni
1      1   34   1
2      2   45   2

    2. Generate random characters of the size of each string

    anonymize_str_vector <- function(x) {
  if( !is.character(x) ) stop('x is not a vector of characters')
  
  anonymize_scalar <- function(s) {
    suppressWarnings(if (!is.na(as.numeric(s))) {sample_data <- 0:9} else {sample_data <- letters})
    paste0(sample(sample_data, nchar(s), replace = TRUE), collapse="")
  }
  sapply(x, anonymize_scalar)
}

anonymize_str_df(df)

  nombre Edad      dni
1   wgon   34 36360598
2  cbval   45 18800229

    Clarification: technically speaking, this form would be reversible if the "seed" with which the process began is known.

    3. Use some "hashing" routine

    Source: Data anonymization in R

    library(digest)

anonymize_str_vector <- function(x, algo="crc32") {
  if( !is.character(x) ) stop('x is not a vector of characters')
  unq_hashes <- vapply(unique(x), function(object) digest(object, algo=algo), FUN.VALUE="", USE.NAMES=TRUE)
  unname(unq_hashes[x])
}

anonymize_str_df(df)

    nombre Edad      dni
1 5c97bc37   34 2670dbda
2 e6473558   45 bc76be3a
    • 2