Create new variable, change character to integer

[CM1]

Hi,

I have a variable called IMD_group with 5 different values (all character, in my tibble here they are factor as I can't work out how to make character in csv and import with datapasta, sorry). The entries are currently messy (i.e. A1 (LQ) should be 1 or one, upwards to 5). Code below doesn't seem to be working. Any ideas?

Thanks

tibble::tribble(
~ID, ~IMD_decile,
1L, "A1 (LQ)",
2L, "A2",
3L, "A3",
4L, "A1 (LQ)",
5L, "A5 (HD)",
6L, "A4",
7L, "A1",
8L, "A4",
9L, "A5 (HD)",
10L, "A2"
)

IMD_one <- c("A1 (LQ)")
IMD_two <- c("A2")
IMD_three <- c("A3")
IMD_four <- c("A4")
IMD_five <- c("A5 (HD)")

tribble1 <- tribble %>%
mutate(IMD_group = case_when(
IMD_quintile %in% IMD_one ~ "2",
IMD_quintile %in% IMD_two ~ "3",
IMD_quintile %in% IMD_three ~ "4",
IMD_quintile %in% IMD_four ~ "5"
))

[michaelbgarcia]

Hi there, it seems you have your variables mixed up - you want to define IMD_quintile using the values in IMD_group. Your code seems to want to do the opposite. Also, you need to define the tribble prior to working on it. Lastly, since you want an integer, I removed the quotes in the recoded values. This creates a dbl vector, so if you need integers you would force that conversion too. How does this look?

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(tibble)

my_tribble = tibble::tribble(
  ~ID, ~IMD_group,
  1L, "A1 (LQ)",
  2L, "A2",
  3L, "A3",
  4L, "A1 (LQ)",
  5L, "A5 (HD)",
  6L, "A4",
  7L, "A1",
  8L, "A4",
  9L, "A5 (HD)",
  10L, "A2"
)

IMD_one <- c("A1 (LQ)")
IMD_two <- c("A2")
IMD_three <- c("A3")
IMD_four <- c("A4")
IMD_five <- c("A5 (HD)")

tribble1 <- my_tribble %>%
  mutate(
    IMD_quintile = case_when(
      IMD_group %in% IMD_one ~ 2,
      IMD_group %in% IMD_two ~ 3,
      IMD_group %in% IMD_three ~ 4,
      IMD_group %in% IMD_four ~ 5
    ),
    IMD_quintile = as.integer(IMD_quintile)
  )

print(tribble1)
#> # A tibble: 10 × 3
#>       ID IMD_group IMD_quintile
#>    <int> <chr>            <int>
#>  1     1 A1 (LQ)              2
#>  2     2 A2                   3
#>  3     3 A3                   4
#>  4     4 A1 (LQ)              2
#>  5     5 A5 (HD)             NA
#>  6     6 A4                   5
#>  7     7 A1                  NA
#>  8     8 A4                   5
#>  9     9 A5 (HD)             NA
#> 10    10 A2                   3

^{Created on 2021-11-29 by the reprex package (v2.0.1)}

[CM1]

Hi,
Thanks very much. It works on my repress but not in the real data frame because I think the A1 (LQ) etc is in character format. It's a secure portal so I can't reprex it. Do you know I can either make a cxv reprex with character variable to add to the thread, or how this code would need to change to make it work for character variable? The error message I get is NAs introduced by coercion - do you know what that means?
Thanks again

[CM1]

I've worked it out in the dummy data with character, will try and apply to real df now. Thanks very much again

[HanOostdijk]

You say it is not working. Do you see that in your end result?

You don't have a rule to translate A5 (HD) : i.e. you do define IMD_five but you don't use that.
So it is not strange that A5 (HD) is matched with NA and that could cause the error message concerning coercion.

[CM1]

Thanks so much to you both. Problem solved!