Conditional assignment of vector in R

Question

I'd like to assign a vecotr based on a conditional statement. How do I do this? This is what I have tried:

zero_rows <- ifelse(i == 1, c(2,3,4,7,12),
               ifelse(i == 2, c(4,7,12),
               ifelse(i == 3, c(2,3,4,7,9,12),
               ifelse(i == 4, c(4,7,9),
               ifelse(i == 5, c(4,12),
               ifelse(i == 6, c(9,12),
               ifelse(i == 7, 12,
               ifelse(i == 8, c(2,3,4,7,9,12), 0))))))))

and

  zero_rows <- case_when(i == 1 ~ c(2,3,4,7,12),
                         i == 2 ~ c(4,7,12),
                         i == 3 ~ c(2,3,4,7,9,12),
                         i == 4 ~ c(4,7,9),
                         i == 5 ~ c(4,12),
                         i == 6 ~ c(9,12),
                         i == 7 ~ 12,
                         i == 8 ~ c(2,3,4,7,9,12),
                         TRUE ~ 0)

Answer 1

It may be a case for switch

fn <- function(i) {
   i <- as.character(i)
   switch(i, 
         "1" = c(2,3,4,7,12),
          "2" = c(4,7,12),
          "3" = c(2,3,4,7,9,12),
          "4" = c(4,7,9),
          "5" = c(4,12),                              
          "6" = c(9,12),
          "7" = 12,
          "8" = c(2,3,4,7,9,12), 
         0)


}

-testing

> fn(1)
[1]  2  3  4  7 12
> fn(2)
[1]  4  7 12
> fn(3)
[1]  2  3  4  7  9 12
> fn(9)
[1] 0
> fn(10)
[1] 0

---

If we want to do this on each value of a column ('col1'), use rowwise and wrap the output in a list

library(dplyr)
df1 %>%
    rowwise %>%
    mutate(out = list(fn(col1)))

Or with map

library(purrr)
df1 %>%
     mutate(out = map(col1, fn))

---

ifelse/if_else/case_when - expects all the arguments to be of same length. Based on the code showed, it is not the case where the length differ based on the input arguments

Answer 2

We can try nested ifelse along with list like below

f <- function(i) {
    ifelse(i == 1, list(c(2, 3, 4, 7, 12)),
        ifelse(i == 2, list(c(4, 7, 12)),
            ifelse(i == 3, list(c(2, 3, 4, 7, 9, 12)),
                ifelse(i == 4, list(c(4, 7, 9)),
                    ifelse(i == 5, list(c(4, 12)),
                        ifelse(i == 6, list(c(9, 12)),
                            ifelse(i == 7, list(12),
                                ifelse(i == 8, list(c(2, 3, 4, 7, 9, 12)), 0)
                            )
                        )
                    )
                )
            )
        )
    )[[1]]
}

Answer 3

It would be most natural to use i as an index, like so:

l <- list(c(2, 3, 4, 7, 12),
          c(4, 7, 12),
          c(2, 3, 4, 7, 9, 12),
          c(4, 7, 9),
          c(4, 12),
          c(9, 12),
          12,
          c(2, 3, 4, 7, 9, 12))

zero_rows <- ifelse(i %in% seq_along(l), l[i], list(0))

Though it would be safer and faster to avoid ifelse altogether:

ok <- i %in% seq_along(l)
zero_rows <- rep.int(list(0), length(i))
zero_rows[ok] <- l[i[ok]]

The first approach assumes that the elements of i are all greater than or equal to 1. The second approach supports any numeric i.