如何从R data.frame中的分类变量采样？

Question

I am trying to sample from an R data frame but I have some problems with the categorical variables. I am not taking a random subsamples of rows but I am generating rows such that the new variables have individually the same distribution of the original one. I am having problem with the categorical variables.

> head(x0)
   Symscore1 Symscore2 exercise3 exerciseduration3                groupchange age3
3          1         0         1                 0 Transitional to Menopausal   52
4          0         0         5                 2 Transitional to Menopausal   62
6          0         0         2                 0 Transitional to Menopausal   54
8          0         0         5                 3 Transitional to Menopausal   56
10         0         0         4                 3 Transitional to Menopausal   59
13         0         1         4                 3 Transitional to Menopausal   55
   packyears     bmi3 education3
3   2.357143 23.24380      Basic
4   2.000000 16.76574 University
6   1.000000 23.30668      Basic
8   1.428571 22.14533 University
10  1.428571 22.14533 University
13  0.000000 22.03857 University
> xa = as.data.frame(sapply(X = x0,  FUN = sample))
> head(xa)
  Symscore1 Symscore2 exercise3 exerciseduration3 groupchange age3 packyears
1         1         0         2                 3           4   49 53.571430
2         0         0         3                 0           3   46  2.142857
3         1         0         3                 3           4   49  4.000000
4         0         1         3                 3           4   58  0.000000
5         0         0         2                 0           1   57  0.000000
6         0         0         3                 0           1   47 26.871429
      bmi3 education3
1 25.84777          2
2 21.25850          2
3 25.79592          3
4 23.93899          1
5 25.97012          2
6 23.53037          2
> X = rbind(x0,xa)
Warning messages:
1: In `[<-.factor`(`*tmp*`, ri, value = c(4, 3, 4, 4, 1, 1, 2, 4, 4,  :
  invalid factor level, NA generated
2: In `[<-.factor`(`*tmp*`, ri, value = c(2, 2, 3, 1, 2, 2, 3, 2, 2,  :
  invalid factor level, NA generated
> 

Answer 1

You could try:

x2 <- x0
x2[] <- lapply(x0,  FUN = sample)

x2
#   Symscore1 Symscore2 exercise3 exerciseduration3                groupchange
#3          0         0         1                 0 Transitional to Menopausal
#4          0         0         5                 3 Transitional to Menopausal
#6          0         0         4                 3 Transitional to Menopausal
#8          0         0         2                 0 Transitional to Menopausal
#10         1         1         4                 3 Transitional to Menopausal
#13         0         0         5                 2 Transitional to Menopausal
    age3
#3    54
#4    59
#6    52
#8    56
#10   62
#13    5

rbind(x0,x2)

data

x0 <- structure(list(Symscore1 = c(1L, 0L, 0L, 0L, 0L, 0L), Symscore2 = c(0L, 
0L, 0L, 0L, 0L, 1L), exercise3 = c(1L, 5L, 2L, 5L, 4L, 4L), exerciseduration3 = c(0L, 
2L, 0L, 3L, 3L, 3L), groupchange = structure(c(1L, 1L, 1L, 1L, 
1L, 1L), .Label = "Transitional to Menopausal", class = "factor"), 
age3 = c(52L, 62L, 54L, 56L, 59L, 5L)), .Names = c("Symscore1", 
"Symscore2", "exercise3", "exerciseduration3", "groupchange", 
"age3"), class = "data.frame", row.names = c("3", "4", "6", "8", 
"10", "13"))