Summarise or mutate + across converts double dbl or integer int to character chr

[sergiouribe]

something curious happens when applying summarise or mutate, where the class of the data changes.

No errors, the outcome class is OK

# Packages
require(NHANES, # the dataset
        tidyverse)


# the dataset
data(NHANES)


# example, no errors
NHANES %>% 
  group_by(Gender) %>% 
  summarise(across(c("Age", "Pulse"), 
                    list(mean, min, max), na.rm = T))

The outcome

I'm trying to make the numbers have one decimal place. So, when

# example, errors, some columns are CHR now
NHANES %>% 
  group_by(Gender) %>% 
  summarise(across(c("Age", "Pulse"), 
                    list(mean, min, max), na.rm = T)) %>% 
  mutate(across(is.numeric, format, 1))

some columns are CHR instead their original class

is this a bug or a feature?

[EconProf]

I believe that format() converts numbers to characters:

> format(123)
[1] "123"
> class(format(123))
[1] "character"

You can use format() to set the decimals and then convert back to numeric:

> as.numeric(format(c(123.4, 123), nsmall = 1))
[1] 123.4 123.0