R dataframe 的更高效的块重采样

Question

I'm trying to resample an R dataframe in a clustered/blocked way. I'm doing so with the code snippet below, but its quite slow:

    index_sample <- sample(unique(data[[cluster_var]]), 
                           size=length(unique(data[[cluster_var]])), replace=T)

    indices <- unlist(sapply(index_sample, FUN=function(x) {which(data[[cluster_var]] == x)}))

Is there a more efficient way to do this? The unlist/sapply step in particular seems very slow.

Example of desired behavior:

set.seed(1919)
data <- data.frame(x=sample(seq(1,5,1), 20, replace=TRUE),
                   y = runif(20))
index_sample <- sample(unique(data[['x']]), 
                       size=length(unique(data[['x']])), replace=T)
indices <- unlist(sapply(index_sample, FUN=function(x) {which(data[['x']] == x)}))
print(indices)
[1]  7  8  9 10 14 17 20  7  8  9 10 14 17 20  1 12  2 18 19  6 11 13 16    

Answer 1

We can use outer

indices2 <- which(outer(data$x, index_sample, FUN = `==`), arr.ind = TRUE)[,1]

-testing with OP's solution

identical(indices, indices2)
#[1] TRUE