Stratified random sampling from a `data.frame` in R

stratified.R

stratified <- function(df, group, size, select = NULL, 
                       replace = FALSE, bothSets = FALSE) {
  if (is.null(select)) {
    df <- df
  } else {
    if (is.null(names(select))) stop("'select' must be a named list")
    if (!all(names(select) %in% names(df)))
      stop("Please verify your 'select' argument")
    temp <- sapply(names(select),
                   function(x) df[[x]] %in% select[[x]])
    df <- df[rowSums(temp) == length(select), ]
  }
  df.interaction <- interaction(df[group], drop = TRUE)
  df.table <- table(df.interaction)
  df.split <- split(df, df.interaction)
  if (length(size) > 1) {
    if (length(size) != length(df.split))
      stop("Number of groups is ", length(df.split),
           " but number of sizes supplied is ", length(size))
    if (is.null(names(size))) {
      n <- setNames(size, names(df.split))
      message(sQuote("size"), " vector entered as:\n\nsize = structure(c(",
              paste(n, collapse = ", "), "),\n.Names = c(",
              paste(shQuote(names(n)), collapse = ", "), ")) \n\n")
    } else {
      ifelse(all(names(size) %in% names(df.split)),
             n <- size[names(df.split)],
             stop("Named vector supplied with names ",
                  paste(names(size), collapse = ", "),
                  "\n but the names for the group levels are ",
                  paste(names(df.split), collapse = ", ")))
    }
  } else if (size < 1) {
    n <- round(df.table * size, digits = 0)
  } else if (size >= 1) {
    if (all(df.table >= size) || isTRUE(replace)) {
      n <- setNames(rep(size, length.out = length(df.split)),
                    names(df.split))
    } else {
      message(
        "Some groups\n---",
        paste(names(df.table[df.table < size]), collapse = ", "),
        "---\ncontain fewer observations",
        " than desired number of samples.\n",
        "All observations have been returned from those groups.")
      n <- c(sapply(df.table[df.table >= size], function(x) x = size),
             df.table[df.table < size])
    }
  }
  temp <- lapply(
    names(df.split),
    function(x) df.split[[x]][sample(df.table[x],
                                     n[x], replace = replace), ])
  set1 <- do.call("rbind", temp)
  
  if (isTRUE(bothSets)) {
    set2 <- df[!rownames(df) %in% rownames(set1), ]
    list(SET1 = set1, SET2 = set2)
  } else {
    set1
  }
}

This is exactly what I was looking for. Thanks so much!

@mattsigal, No problem--glad it helped. Also, check out the data.table version if you need something speedier.

Thanks. This is great!

Dearr mrdwab
Hope this find you well,
really it great code,
and i want to ask you question please about
stacked regression as it mention in article of Brieman
http://statistics.berkeley.edu/sites/default/files/tech-reports/367.pdf

have you any code for that, please.

This is genius. Thanks!

Thank you for posting this! It's just what I needed! I am using this function to prepare some data for my research. Do you have a preferred citation so I can cite your function in my paper?

Hi there! Great function! I am using it to randomly select a subset of species observations so that we can verify the species identification of a randomly selected subset. So, I can easily stratify by species. However, if I want the observations to also be stratified by transect so that if possible, the species checked are from different transects, it becomes more complicated. I tried: size=c("TransectName"=1,"SpeciesName"=5)
to choose 5 observations of each species, each from a different transect, but this didn't work.
Error message: (Error in stratified(df, group = c("SpeciesName", "TransectName"), size = c(TransectName = 1, : Number of groups is 508 but number of sizes supplied is 2)
It gets more complicated, because if it isn't possible to get 5 observations from different transects, they can come from the same transect.
Any ideas on how I would accomplish this?
If not, I'll drop the stratification by transect.
Thanks!!

Great function! I have a case in which I am not sure it could be used.

Let's say I have a dataset x and a dataset y. Dataset x contains N observations and dataset y contains M observations, where N>M. Both datasets contain the same variables k. If I want to make from dataset x a representative sample of dataset y.
Is it possible to use the function and to specify the proportions of dataset y for the category vars of: 1) Size and 2) Sector?

stratified(Dataset x, c("Size", "Sector"), ...)

Thanks very much!

Great Function! @mrdwab Could you please provide a official citation guide to cite your function/package? Thank you!

thank you!

Great Function !!!!!......Thanks a lot !!

Great. Thank you very much for that.

awesome!

Hi, Thank you for the amazing code. But i have a query regarding using multiple columns to create strata.
Here you have shown one example "stratified(dat1, c("E", "D"), size = 0.15)" where both "E" and "D" are categorical columns. I was wondering if we can use multiple numerical columns. Please guide me for the same.
Basically your code : stratified(dat1, c("B", "C"), size = 0.15) should return some output.
Thanks in advance.

Hi, I tried to load the function using the following commands:

library(devtools)
source_gist("https://gist.github.com/mrdwab/6424112")

But, I got the following error:

Error in r_files[[which]] : invalid subscript type 'closure'

Really appreciate your help to fix this. This is exactly the function that I have been looking for and desperately need to use it.

Wow this is exactly what I need! Thank you so much!

By the way, is there a way to apply population weights for the sampling?

Thanks so much for this code, it works perfectly.

Hi there Ananda,

How do I make attribution to your article?
Such as citing the material. This is top stuff, indeed.