使用dplyr mutate在组中查找第一次出现的值

Question

How do i find the first occurence of a certain value, within a group using dplyr .

The following code gives the desired result, but it I'm wondering if there is a shorter way to do it.

Also, I am worried that group_by or mutate , or some other function might do implicit rearrangement of the rows, don't know if this could be an issue?

mtcars   %>% select( cyl, carb) %>% group_by( cyl ) %>%

   mutate( "occurence_of_4" =  carb == 4 )  %>%

   dplyr::arrange( cyl )  %>%

   group_by( cyl, occurence_of_4)  %>%

   mutate( "count" = 1:n(),
           "first_4_in_cyl_group"  = ifelse( occurence_of_4==TRUE & count==1, TRUE, FALSE)) 

The variable first_4_in_cyl_group is TRUE for the first occurence of "4" in each cylinder group, FALSE otherwise:

Source: local data frame [32 x 5]
Groups: cyl, occurence_of_4

   cyl carb occurence_of_4 count first_4_in_cyl_group
1    4    1          FALSE     1                FALSE
2    4    2          FALSE     2                FALSE
3    4    2          FALSE     3                FALSE
4    4    1          FALSE     4                FALSE
5    4    2          FALSE     5                FALSE
6    4    1          FALSE     6                FALSE
7    4    1          FALSE     7                FALSE
8    4    1          FALSE     8                FALSE
9    4    2          FALSE     9                FALSE
10   4    2          FALSE    10                FALSE
11   4    2          FALSE    11                FALSE
12   6    4           TRUE     1                 TRUE
13   6    4           TRUE     2                FALSE
14   6    1          FALSE     1                FALSE
15   6    1          FALSE     2                FALSE
16   6    4           TRUE     3                FALSE
17   6    4           TRUE     4                FALSE
18   6    6          FALSE     3                FALSE
19   8    2          FALSE     1                FALSE
20   8    4           TRUE     1                 TRUE
21   8    3          FALSE     2                FALSE
22   8    3          FALSE     3                FALSE
23   8    3          FALSE     4                FALSE
24   8    4           TRUE     2                FALSE
25   8    4           TRUE     3                FALSE
26   8    4           TRUE     4                FALSE
27   8    2          FALSE     5                FALSE
28   8    2          FALSE     6                FALSE
29   8    4           TRUE     5                FALSE
30   8    2          FALSE     7                FALSE
31   8    4           TRUE     6                FALSE
32   8    8          FALSE     8                FALSE

Answer 1

You may use !duplicated .

mtcars %>%
  select(cyl, carb) %>%
  group_by(cyl) %>%
  mutate(first_4 = carb == 4 & !duplicated(carb == 4))  %>%
  arrange(cyl)

Answer 2

A couple of modifications:

1. Remove the first mutate step by creating the "occurence_of_4" variable within the group_by

2. ifelse is not needed as the output will be "TRUE/FALSE"

    library(dplyr) mtcars %>% select(cyl, carb) %>% group_by(cyl, occurence_of_4= carb==4) %>% arrange(cyl) %>% mutate(count= row_number(), first_4_in_cyl_group = occurence_of_4 & count==1) 

Answer 3

Instead of grouping it is enough to arrange by cyl and carb. With lag you can check the previous value.

mtcars database doesn't have an ID column, so if you are rearranging rows you could add them with add_rownames (as docendodiscimus suggested in comments).

mtcars   %>% 
  select( cyl, carb ) %>%
  add_rownames() %>%
  arrange(cyl, carb) %>%
  mutate(
    isfirst = (carb == 4 & (is.na(lag(carb)) | lag(carb) != 4))) %>%
  filter(isfirst)

The result:

#      rowname cyl carb isfirst
# 1  Mazda RX4   6    4    TRUE
# 2 Duster 360   8    4    TRUE