What parameter must an R function have to use it within the mutate function from tidyverse?

Question

A have a tibble with a column with strings representing hours and minutes. I want to tidy that column and convert the elements into integers representing just the minutes.

Those strings can have one of the following forms:

"5" (which means 5 minutes)
"XX min" (meaning xx minutes)
"X Std" (meaning x hours)
"X Std. YY min" (meaning x hours and yy minutes)

I wrote a function to convert those strings into minutes.

"5" should become 5.
"45 min" should become 45.
"2 Std" should become 120.
"1 Std. 30 min" should become 90.

This is what the function looks like:

convert_ZA_time <- function(string) {
    if (nchar(string) == 1) {
      result <- as.integer(string)
    }
    else if (endsWith(string, " Std")) {
      result <- as.integer(substring(string, 1, 1)) * 60
    }
    else if (endsWith(string, " min") && nchar(string) == 6) {
      result <- as.integer(substring(string, 1, 2))
    }
    else if (endsWith(string, " min") && nchar(string) > 6) {
      hour <- as.integer(gsub(" Std.*", "", string, perl = TRUE))
      minute_plus <- gsub("^\\d+ Std. ", "", string, perl = TRUE)
      minute <- as.integer(gsub(" min$", "", minute_plus))
      result <- hour * 60 + minute
    }
    else {result <- NA}
    return(result)
}

Testing with strings it works just fine:

convert_ZA_time("2 Std. 50 min")
# prints [1] 170

But when I try to use this function inside the tidyverse mutate function I get the following error:

df <- tibble(datestr = c("5", "45 min", "1 Std", "2 Std. 30 min"))
df2 <- df %>% mutate(minutes = convert_ZA_time(datestr))
# throws error: the condition has length > 1 and only the first element will be used

How do I have to change my function to use it within mutate correctly?

P.S. as I understand it: mutate takes every "datestr" and puts it into the function "convert_ZA_time". But apparently mutate puts a vector into the function?

Thanks for any help!

jay.sf jay.sf · Accepted Answer · 2020-10-06T11:36:52

Your function just isn't Vectorized yet.

convert_ZA_time(c("2 Std. 50 min", "3 Std. 50 min"))
# [1] 170 230
# Warning messages:
# 1: In if (nchar(string) == 1) { :
#   the condition has length > 1 and only the first element will be used
# 2: In if (endsWith(string, " Std")) { :
#   the condition has length > 1 and only the first element will be used

Fix:

convert_ZA_timev <- Vectorize(convert_ZA_time)
      
convert_ZA_timev(c("2 Std. 50 min", "3 Std. 50 min"))
# 2 Std. 50 min 3 Std. 50 min 
#           170           230

Explanation

You have an if / else structure in your function, like this one:

fun <- function(x) if (x >= 0) "pos" else "neg"

When applied to a vector with length greater than one, it evaluates just the first element with a warning.

v <- -2:2

fun(v)
# [1] "neg"
# Warning message:
#   In if (x >= 0) "pos" else "neg" :
#   the condition has length > 1 and only the first element will be used

fun(v[1])
# [1] "neg"

Vectorization enables a function to deal with vectors.

funv <- Vectorize(fun)
funv(v)
# [1] "neg" "neg" "pos" "pos" "pos"

What parameter must an R function have to use it within the mutate function from tidyverse?

1 Answers

Explanation