R function to recode values or similar (like SPSS)-CodePudding

I have a dataframe with this structure:

id v1 v2 v3 v4 v5
1  .  1  .  2  3
2  1  2  3  .  .
3  3  2  .  1  .

And I want to recode it like this

id v1 v2 v3 v4 v5
1  0  3  0  2  1
2  3  2  1  0  0
3  1  2  0  3  0

So I want to recode values (1=3) (missing = 0) (3=1)

In SPSS this is fairly intuitive:

recode v1 to v5 (sysmis=0)(3=1)(1=3)(else=copy).

I'm looking for a similar approach in R. No preference on either base or tidyverse aproach.

CodePudding user response：

Replace the . to NA, convert the type of columns (type.convert) to numeric, subtract from 4 and replace the NA with 0

df[df == "."] <- NA
df <- type.convert(df, as.is = TRUE)
df[-1] <- 4- df[-1]
df[is.na(df)] <- 0

Or using tidyverse

library(dplyr)
library(tidyr)
df %>% 
  mutate(across(-id, ~ replace_na(4 - as.numeric(.x), 0)))

-output

  id v1 v2 v3 v4 v5
1  1  0  3  0  2  1
2  2  3  2  1  0  0
3  3  1  2  0  3  0

-output

> df
  id v1 v2 v3 v4 v5
1  1  0  3  0  2  1
2  2  3  2  1  0  0
3  3  1  2  0  3  0

data

df <- structure(list(id = 1:3, v1 = c(".", "1", "3"), v2 = c(1L, 2L, 

2L), v3 = c(".", "3", "."), v4 = c("2", ".", "1"), v5 = c("3", 
".", ".")), class = "data.frame", row.names = c(NA, -3L))

CodePudding user response：

This was surprisingly difficult with recode, but it is I think a more readable answer:

library(dplyr)
dat %>% 
  mutate(across(v1:v5, ~ .x %>% 
                  replace(. == ".", 0) %>% 
                  as.numeric() %>% 
                  recode(`3` = 1, `1` = 3)
                  ))

  id v1 v2 v3 v4 v5
1  1  0  3  0  2  1
2  2  3  2  1  0  0
3  3  1  2  0  3  0