CODEWITHSUNDEEP
rdplyr
Damien Dotta
Damien Dotta

Reputation: 939

How the get this expected result with dplyr

From my tibble test, I would like to get result :

test <- tribble(
  ~colA, ~regsiege, ~nbeta_reg52, ~nbeta_reg53, ~nbeta_reg75,
  "a",   "52", 0,1,3,
  "b",   "53", 0,1,2,
  "c",   "75", 3,2,1
)

# expected tibble
result <- tribble(
  ~colA, ~regsiege, ~nbeta_reg52, ~nbeta_reg53, ~nbeta_reg75,
  "a",   "75", 0,1,3,
  "b",   "53", 0,1,2,
  "c",   "53", 3,2,1
)

I tried something like this with this condition :

for maregion in c("52","53","75"){
  test2 <- test %>% 
    mutate(
      regsiege=if_else(regsiege==maregion & !!sym(paste0("nbeta_reg",maregion))==0,
                       ...)
    )
}

EDIT :
I would like to replace the values ​​in regsiege column based on this rule: if regsiege = "XX" and nbeta_regXX == 0 then regsiege == "XX" ie the suffix of the column in which nbeta_regXX is the maximum

Upvotes: 0

Views: 31

Answers (1)

r2evans
r2evans

Reputation: 160447

Here's a pipeline that implements what you say, though not what you have in result:

library(dplyr)
library(tidyr) # pivot_longer, pivot_wider

test %>%
  pivot_longer(
    starts_with("nbeta"), names_pattern = "(.*[^0-9])([0-9]+)", 
    names_to = c("nbeta_reg", "num")
  ) %>%
  group_by(colA) %>%
  mutate(
    regsiege = if (any(regsiege == num & value == 0)) num[which.max(value)] else regsiege,
    nbeta_reg = paste0(nbeta_reg, num)
  ) %>%
  select(-num) %>%
  ungroup() %>%
  pivot_wider(colA:regsiege, names_from = "nbeta_reg", values_from = "value")
# # A tibble: 3 x 5
#   colA  regsiege nbeta_reg52 nbeta_reg53 nbeta_reg75
#   <chr> <chr>          <dbl>       <dbl>       <dbl>
# 1 a     75                 0           1           3
# 2 b     53                 0           1           2
# 3 c     75                 3           2           1

The reason it is different is that your rule says that if nbeta_regXX == 0, but in the third row, nbeta_reg75 is not 0, so regsiege should be unchanged.

Upvotes: 1

Related Questions