Package: base

Function: round()

1. Round our score values

Review the data (d1)

# A tibble: 6 x 3
     id scorex scorey
  <dbl>  <dbl>  <dbl>
1    30    3.6 101.51
2    34    8.5 132.65
3    35    5.5 155.65
4    36    5.3 129.75
5    37   NA   109.6 
6    38    4.2 115.31

Round our score values using base::round().

  • Note: We need to use dplyr::mutate() in order to modify our existing columns.
  • Note: We use dplyr::across() to mutate both scorex and scorey with the same function.
d1 %>% 
  dplyr::mutate(dplyr::across(scorex:scorey, ~ round(.)))
# A tibble: 6 x 3
     id scorex scorey
  <dbl>  <dbl>  <dbl>
1    30      4    102
2    34      8    133
3    35      6    156
4    36      5    130
5    37     NA    110
6    38      4    115

There are a few things to note here.

First, the default is to round to a whole number. We can add the argument digits to change the number of digits that we want.

Second, the rounding behavior may not be as expected. This base R function uses banker’s rounding. This means that R rounds to the even number when rounding at 5. So for example, when you have a value of 23.5, it will round up to 24. When you have a value of 22.5, it will round down to 22. See documentation on round for more information.

Let’s see what happens when we round to 1 digit.

d1 %>% 
  dplyr::mutate(dplyr::across(scorex:scorey, ~ round(., digits = 1)))
# A tibble: 6 x 3
     id scorex scorey
  <dbl>  <dbl>  <dbl>
1    30    3.6  101.5
2    34    8.5  132.7
3    35    5.5  155.7
4    36    5.3  129.8
5    37   NA    109.6
6    38    4.2  115.3

In this case it appears that rounding is doing what we would expect, rounding up at 5, no matter if the tens digit is odd or even.

However, it is important to note that when you round with 1 or more digits, since decimal places are not represented exactly, you may get different results depending on your OS: 1.75 may round to 1.7 or 1.8.

  • Note: The other options in base R for rounding are base::ceiling() if you want to always round up, base::floor() if you want to always round down, and base::trunc() if you want to simply cut off all decimal places and keep the current whole value.


Package: janitor

Function: round_half_up()

1. Round our score values

Review the data (d1)

# A tibble: 6 x 3
     id scorex scorey
  <dbl>  <dbl>  <dbl>
1    30    3.6 101.51
2    34    8.5 132.65
3    35    5.5 155.65
4    36    5.3 129.75
5    37   NA   109.6 
6    38    4.2 115.31

Round our score values.

d1 %>%
  dplyr::mutate(dplyr::across(scorex:scorey, ~ janitor::round_half_up(.)))
# A tibble: 6 x 3
     id scorex scorey
  <dbl>  <dbl>  <dbl>
1    30      4    102
2    34      9    133
3    35      6    156
4    36      5    130
5    37     NA    110
6    38      4    115

You can see that now we have the behavior that we would normally expect, where round up at 5 or higher and down at 4 or lower.

Let’s see how this function works when we round to 1 digit.

d1 %>%
  dplyr::mutate(dplyr::across(scorex:scorey, ~ janitor::round_half_up(., digits = 1)))
# A tibble: 6 x 3
     id scorex scorey
  <dbl>  <dbl>  <dbl>
1    30    3.6  101.5
2    34    8.5  132.7
3    35    5.5  155.7
4    36    5.3  129.8
5    37   NA    109.6
6    38    4.2  115.3

Again, the rounding here is as expected. However, as mentioned above, this may differ slightly based on operating system. I am currently using Windows 10.

Return to Rounding