paste0 within a for loop in R

Question

What I'd like to obtain is a dataset like this:

ID var
1  t0
1  t1
1  t2
2  t0
2  t1
2  t2

where t restart from 0 for each ID.

If I use:

all <- data.frame(ID=character(),var=numeric())

for (i in 1:2) {
  for (j in 0:3) {
    df <- data.frame(matrix(c(rep(i,3),paste0("t",j)), byrow = F, ncol = 2, nrow = 6))
    all <- rbind(all,df) 
  }
}

there is something clearly wrong. How can I manage it?

you don't need the inner-loop or the matrix to dataframe cohercion Try this: `j = 0:2; for (i in 1:2) { df <- data.frame(ID = rep(i,3), var = paste0("t",j)) all <- rbind(all,df) };` — Nate, Feb 08 '21 at 14:14

G. Grothendieck · Answer 1 · 2021-02-08T14:44:32.060

1) Change the line marked ## as shown.

all <- data.frame(ID=character(),var=numeric())

for (i in 1:2) {
  for (j in 0:3) {
    df <- data.frame(ID = i, var = paste0("t", j))  ##
    all <- rbind(all,df) 
  }
}

2) or use expand.grid

expand.grid(var = paste0("t", 0:3), ID = 1:2)[2:1]

3) Another possibility is crossing from the tidyr package:

library(tidyr)
crossing(ID = 1:2, var = paste0("t", 0:3))

4) there are a number of packages that support python and haskell like comprehensions (comprehenr, eList, listcompr). Using the first:

library(comprehenr)
to_df(for(i in 1:2) for(j in 0:3) list(ID = i, var = paste0("t", j)))

Also note comments below this answer for an additional option.

One more is `CJ(ID = 1:2, var = paste0("t", 0:3))` from `data.table`. — s_baldur, Feb 08 '21 at 14:39

score 0 · Answer 2 · answered Feb 08 '21 at 14:15

I recommend against building a data.frame iteratively like that: it works as expected initially, but as it grows larger, it starts to get much slower, quickly (because every time you add a row, it makes a complete copy of all rows). Better to form a list of frames, and then combine them once. See https://stackoverflow.com/a/24376207/3358227.

To build your frame, here are a couple of ways.

Build it that way from the start, all are groups of three:

data.frame(ID = rep(seq_len(3), each = 3), var = paste0("t", rep(seq_len(3) - 1, times = 3)))
#   ID var
# 1  1  t0
# 2  1  t1
# 3  1  t2
# 4  2  t0
# 5  2  t1
# 6  2  t2
# 7  3  t0
# 8  3  t1
# 9  3  t2

If your data already exists and you just want to add the var field:

dat <- data.frame(ID = rep(seq_len(3), each = 3))
dat$var <- paste0("t", ave(dat$ID, dat$ID, FUN = seq_along) - 1)

Thanks, I'd be faster but in my real data the ID variable it is not just 1,2,etc, so I need the foor loop at least for the i. — jeff, Feb 08 '21 at 14:32
In your real data, you build a `data.frame` from scratch with arbitrary numbers? If you're doing it in a `for` loop, then you must know the IDs and each length beforehand, in which case you don't need a `for` loop. — r2evans, Feb 08 '21 at 15:02

score 0 · Answer 3 · answered Jun 12 '21 at 06:37

Using the newest version (0.3.0) of the listcompr package, you can use {} brackets within characters as placeholders for variables. For instance "t{j}" is evaluated to "t1", "t2", ... for j = 1, 2, ....

Your example data set can be created using gen.data.frame from the listcompr package:

> library(listcompr)
> gen.data.frame(c(ID = i, var = "t{j}"), j = 0:2, i = 1:2)
  ID var
1  1  t0
2  1  t1
3  1  t2
4  2  t0
5  2  t1
6  2  t2

paste0 within a for loop in R

3 Answers3