Arranging rows in custom order using dplyr

Question

With arrange function in dplyr, we can arrange row in ascending or descending order. Wonder how to arrange rows in custom order. Please see MWE.

Reg <- rep(LETTERS[1:3], each = 2)
Res <- rep(c("Urban", "Rural"), times = 3)
set.seed(12345)
Pop <- rpois(n = 6, lambda = 500000)
df <- data.frame(Reg, Res, Pop)

df
   Reg   Res    Pop
1    A Urban 500414
2    A Rural 500501
3    B Urban 499922
4    B Rural 500016
5    C Urban 501638
6    C Rural 499274

df %>%
  arrange()

Desired Output

   Reg   Res    Pop
5    C Urban 501638
6    C Rural 499274
1    A Urban 500414
2    A Rural 500501
3    B Urban 499922
4    B Rural 500016

akrun · Accepted Answer · 2017-09-09T10:49:14.917

137

We can use factor to change the order in a custom way

df %>%
    arrange(factor(Reg, levels = LETTERS[c(3, 1, 2)]), desc(Res), desc(Pop))
#  Reg   Res    Pop
#1   C Urban 501638
#2   C Rural 499274
#3   A Urban 500414
#4   A Rural 500501
#5   B Urban 499922
#6   B Rural 500016

Or with match to get the index and arrange on it

df %>%
    arrange(match(Reg, c("C", "A", "B")), desc(Res), desc(Pop))

If we have multiple columns to arrange in descending order

df %>%
     arrange_at(2:3, desc) %>%
     arrange(match(Reg, c("C", "A", "B")))

edited Sep 09 '17 at 10:49

answered Sep 09 '17 at 09:49

akrun

874,273
37
540
662

2

Thanks @akrun for nice solution. Is there any was to use `desc` for all remaining variables without typing their names? Thanks – MYaseen208 Sep 09 '17 at 10:28
@MYaseen208 Thanks for the note. I added another option. Hope it helps – akrun Sep 09 '17 at 10:49
Yes, to echo @MYaseen208, is there any way to pick a couple to be at the bottom, but have the rest at the top? For instance, if i have 20 vars, and i want 2 of them at the bottom, this method seems to only work if I name the first 18 in the match. Regardless of the desc() and !match I use, I always get the unnamed values at the bottom. – jzadra Feb 22 '18 at 21:11
1

Just a side note for future viewers: first solution can be shortened to `df %>% arrange(factor(Reg, levels = LETTERS[c(3,1,2)]))` – Jan 13 '19 at 16:31
2

It seems that if the sorting variable is already defined as an ordered factor outside of `arrange`, the behaviour of `arrange` is still to order alphabetically... How come? – Fons MA Apr 07 '19 at 23:50
@FonsMA Not clear about the comment. Can you post a small example – akrun Apr 08 '19 at 03:18

score 14 · Answer 2 · answered Dec 17 '20 at 12:32

14

I used the slice() function:

   df %<>%
   slice(5,6,1:4)

answered Dec 17 '20 at 12:32

Peter Dorey

141
1
2

1

I found this the simplest and easiest to follow answer but found I could use %>% rather than %<>% – Markm0705 May 19 '22 at 12:43

score 4 · Answer 3 · answered Sep 09 '17 at 12:20

4

using data.table

df1[order(factor(Reg,levels = LETTERS[c(3,1,2)]))]

Here df1 is data.table

answered Sep 09 '17 at 12:20

dondapati

829
6
18

Arranging rows in custom order using dplyr

3 Answers3

Linked

Related